WO2008071454A2

WO2008071454A2 - Method and arrangement for processing ultrasonic image volumes as well as a corresponding computer program and a corresponding computer-readable storage medium

Info

Publication number: WO2008071454A2
Application number: PCT/EP2007/011211
Authority: WO
Inventors: Harald Reindell
Original assignee: Unbekannte Erben Nach Harald Reindell, Vertreten Durch Den Nachlasspfleger, Rechtsanwalt Und Notar Pohl, Kay-Thomas; Pohl, Kay-Thomas
Priority date: 2006-12-12
Filing date: 2007-12-12
Publication date: 2008-06-19
Also published as: WO2008071454A9; WO2008071454A3

Abstract

The invention relates to a method and an arrangement for processing ultrasonic image volumes as well as to a corresponding computer program and a corresponding computer-readable storage medium. In the method for processing ultrasonic image volumes, the invention proposes using a transducer of a 3D ultrasound scanner to detect the 3D ultrasonic image volumes, detecting the position and orientation of the transducer while detecting the 3D ultrasonic image volumes, and combining them with the respective SD ultrasonic image volumes detected. Provision is also made for detected SD ultrasonic image volumes to be combined, while evaluating the respective position and orientation, with other detected 3D ultrasonic image volumes to form an extended ultrasonic image volume (meta volume) and to be stored, and/or for the detected 3D ultrasonic image volumes to be stored individually together with the combined data relating to the position and orientation.

Description

Method and arrangement for processing ultrasound image volumes and a corresponding computer program and a corresponding computer-readable storage medium

description

Multi-modal method for detection and restitution of largely artifact-free, resolution-preserving and relevant physiological factors taking into account static and dynamic ultrasound image volumes ^* - which can exceed the size of the sound field or sound space of a transducer ^** - and their use in visualization and simulation as an off-line or online acting system.

Introduction: conceptual, basic and environmental understanding

In the following an attempt is made to convey definitions of the most important technical terms and their relevance primarily with regard to the patent application, but also sketching the development of this entire discipline in order to make the connections and dynamics of this development somewhat comprehensible - and to create a few first "lookup" pages for further work if needed.

Definitions of terms and / or explanations are given in bold - and apart from the italics in formulas: Notes on specific need for discussion on the patent application or for internal use only Text parts that are commenting on it are presented here / V.

The figures are as follows:

Fig. 1: B-mode transducer with convex sound field;

* The order of the mentioned characteristics is arbitrary and no significance order

** "Transducer sound field / sound space" refers to a static scan position - it can be a 2D, 3D, 4D or higher dimensional transducers / scanners as described in HUMR sub-procedures specified. Fig. 2: Volume Transducer with "Tilting Scan Field;

Fig. 3: Matrix-Transducer for fast "Real-Time 4D";

Fig. 4: B-mode transducer during a manual sweep with a given sweep speed for approximate volume reconstruction (with a "twisted" transducer at approx. 90 °, but also the same sweep direction often for generation

"Juxtaposed" assembled "panoramic 2D images" performed);

Fig. 5: Manual sweep with B-mode transducer + 3D sensor for basically geometrically correct volume reconstruction (as part of the SD sensor accuracy) from "2D" images arranged one behind the other; Fig. 6: General structure of a 3D upgrade system and conversion scheme 2D to 3D:

Fig. 7: Transformation of individual B-images into a regular volume raster (Cartesian system);

Fig. 8: Above: electromagnetic 3D tracker - top left: a complete sensor system, top right: the receiver mounted on the transducer, below: setup with mobile scanner during a "sweep";

Fig. 9: Individual B-images and their associated coordinates in 3D space at or immediately after detection and before geometric transformation (see Fig. 6). Bottom right: 3D after transformation; Fig. 10: Audio signal of the fetal doppler component of the trigger system f. Image acquisition;

Fig. 11: GUI of the acquisition system for triggered acquisition f 4D volumes;

Fig. 12: Scheme of the processing sequence: Freehand acquisition with FET / ECG

Trigger for 4D reconstruction; Fig. 13: Entire device setup FET during acquisition, including trigger signal graphics.

Ultrasound (diagnostics) is the most general term for the known medical imaging (diagnostic) method used herein.

Ultrasonography is often used as a synonym for ultrasound, but cardiologists have introduced echocardiography conceptually, which actually also has a certain inherent autonomy in terms of device technology, since the gray scale resolution / differentiation (eg a cardinal quality criterion in abdominal ultrasound) is somewhat less important , while here, due to the representation of (fast) moving structures (heart, large Vessels), the temporal resolution (number of images or volumes per sec.) Plays a significant role.

The largest department of ultrasonography is "gynecology and obstetrics" (Ob / Gyn), with regard to "sonography" the latter more appropriately described as "prenatal medicine." Internal medicine (gastro-enterology, endocrinology, etc.) is also important for sonography and next largest There are no further listings - today there is hardly any department that does not use ultrasound (sonography / echocardiography), because not everyone adheres strictly to the above differentiation, one often has to deduce from the context whether "sonography" incl exclusive echocardiography is understood.

Frames (English) are pictures, by which one understands also German commonly 2D-pictures, conceptually but in the typical use probably not quite as compelling as with the English term "frames".

Volumes are here 3D images, so-called volumes, which also well-moving volumes are called logical: A dynamic volume, referred to as 4D, is actually only seen as a series of related (3D) volumes.

Pixels are pixels in a digital 2D image (frame). Voxels are pixels in a digital 3D image (volume).

Scanner is (here) a common (English) term for "ultrasound device" (so German actually most common term) - in the narrower sense, the signal-processing and

-visualiserende central unit, usually under "scanner" but the entirety of the

System with the (on the patient) signal generating "Transducer" (s.u.), the

Sound waves emitted and received, understood - so here, if the transducer is not explicitly mentioned in addition. Signal processing from the transducer - i. According to the native radio frequency or "raw signal" - internally scanners nowadays are practically only digital, which for the internal computer technology currently means almost only "PC-based".

On the one hand, this indicates a price erosion of scanners (now "double-tracked" by the manufacturers - ie still relatively expensive high-end product lines with ever more powerful new features (which are only a few reasonably complete). yours), as well as low-end and mobile product lines for "everyone" and new mobile device markets - answered), on the other hand to a - even increasingly affordable - at least partial medical paradigm shift, so to speak: "from stethoscope to sonoscope". Both, but in particular the second tendency in scanners, among other favorable factors, have a (particularly practical!) Training requirement - and ultrasound has long been the most widespread imaging method in the industry, even before the "PC-based generations" Despite the increasingly competitive competition of alternative radiological procedures such as X-ray, CT, MR, etc., the ultrasound will maintain a strong position - it is radiation-free, even the slight warming of sonicated body regions is recognized as innocuous, and in its mobility and cost-effectiveness is unbeatable in the long run, and its performance depends first and foremost on a "cardinal parameter": the qualification of its user.) The basic principle of such scanners is the well-known propagation speed of sound waves into water (with the interior of the body, mostly water is, about equable) or their reflection at (smallest to large) interfaces and a conversion of sound intensity signals into image signals. Water or a high water content is a prerequisite for the applicability of the method - thus air is not suitable here as a sound conductor, even bones have a low water content - in short, therefore, ultrasound is suitable as a "diagnostic method for more or less soft tissue". Apart from lungs and skeletons, however, they are responsible for the vast majority of the human body (gases, eg in the intestine, may also be disturbing in ultrasound.) The good suitability of water as a sound conductor is primarily due to its low attenuation , Scanners used the speed of sound in water, but is not completely correct - here are the facts written off: speed soft tissue / fluid tissue: 1500/1600 m / sec, air: 355 m / seα, bone: 3500 m / sec. and attempting to calculate this at body temperature and normal air pressure), and scanners are calibrated to a soft tissue mean of 1540 m / sec. which is slightly (approx. 2%) should be above pure water at body temperature - very undramatic, but the point of reference is (so) just a "medically based definition" and not water.

Scanners generally also have a (VHS or S-VHS) output (socket) for connecting a video recorder for image storage or "documentation." Data was also collected here for image (post) processing via Frame Grabber (PCI card) can be gitalisiert and processed in a computer then so - in some cases still in use, but in the long term hardly seminal and technically alone because of the (scanner FrameGrabber :) D / A - A / D - conversion not exactly elegant. After all - the data collected in this way can then be processed practically any desired - in the case of 3D conversion (as well as simulation) together with the associated location data of the 3D sensor, and for 4D (as well as simulation) including the assigned data of one (eg ECG -) trigger. This method has been used (here: for the simulator) in 2D scanners - why it because of a "unison obstruction policy" of the (3D) scanner manufacturers: "Everything is secret, even our data format" - and hardly alternatives gave. In the last two years, however, it has been possible to rethink the "global players" of ultrasound - Hitachi, Medison / SonoAce, GE (General Electric), Philips and others - to rethink what was urgently needed: the data acquisition of native 3D / 4D data sets is only possible digitally - technically more "elegant" overall and desirable for image quality alone. Currently (see below 3D / 4D) the "production process for the simulation" is still "double-tracked" - ie "conventional" method (acquisition: 2D scanner => computer + FrameGrabber + 3D sensor / processing => 3D / 4D => Simulation: 2D sections through 3D / 4D), which currently has some partial strengths (especially the size of the volumes that can be generated) on the one hand - and the digital data transfer of native 3D / 4D data sets for the subsequent simulation on the other - practiced.

(However, a major shortcoming of 3D / 4D datasets scanned to date has been their small size, which was limited to a single 3D / 4D transducer position - a process for "sticking together" native volumes in a dataset - analogous to a "sweep" Capturing 2D to map larger anatomical areas or volumes did not exist - at least not in the geometric accuracy required for simulation / training and as an "importable dataset", from other simulation-relevant aspects - more about that HUMR - not to mention. ) Although this potentially creates a certain dependency, on the other hand, the scanner manufacturers now also consider it important to be represented in the training via their systems or images captured there (which is almost a "competitive situation"). In addition, hospitals also wish for their digital data storage - required for archiving, documentation, etc. - (eg the increasingly important, medical-specific DICOM standard, which also has its own image format - however, as specific as "pizza", of which there are "a hundred" variants because proprietary objects also exist) - increasingly disclosure. The DICOM standard (for digital imaging and communications in medians) is thus an open standard for the exchange of images in medicine. The current version is PS 3 2006. The DICOM dataset serves as a container and contains not only one or more images but also meta information such as patient name, date of admission, device parameters or doctor's name, and is also successively extended with curve, radiation therapy data and much more. The visualization takes place on so-called PACS systems (for Picture Archival and Communication System). In general, manufacturers who manufacture medical devices and use DICOM must submit a descriptive document, the so-called Conformance Statement, for their products. All of this is a very slow but long-term and unstoppable trend. Theoretically, it could be a competitive risk for a simulator manufacturer that cases would then be available from many hospitals arbitrarily for simulation, in fact - and this already makes this patent application itself quite clear - it needs for a good simulator much more than just "Digital ultrasound cases" (in hospitals so far often stored only as a set of diagnostically "relevant" 2D ergrebn / s sections), which just were not specifically recorded and edited for the simulation. A reversal of this openness or greater risks from this openness are therefore unlikely for the foreseeable future. However, one could at least feed and produce at least "bad simulators" with such data sets, so it is also a strategic task of this patent, for this area of medical quality control and assurance (here: ultrasound diagnosis or -Education / Qualification) to create and establish quasi a quality standard (there is also consensus with virtually all experts in this field in medicine), then later not a market with bad products and resulting poor (training) quality cumbersome To clean it up, but to prevent it in time - as it were paradigmatic - by a reasonably high standard as possible. The "pitfalls" of the ultrasound simulation and its solutions show (hopefully) self-explanatory that on the one hand this is actually necessary and desirable (these solutions are intended to substantially - partly drastically: - prevent false and misleading diagnostic images or faulty / deficient training) and on the other hand, technically feasible with justifiable effort, as (ir) relevant for granting a patent may be - these are desirable additional effects in the matter. Transducer is the English term for the measurement-generating "transducer of an ultrasound device, which the examiner (usually on the outer skin, but there are also other, so-called" invasive "procedures) of the patient touches down - there are several synonyms such as: Probe, German probe etc. To transmit the "raw data" generated by the transducer, transducers are connected to the scanner with a "proprietary" cable.

There are usually several (also interchangeable with the individual scanners) - with different "penetration depths" or frequencies - the lower the frequency, the higher the penetration depth, and the lower the geometric resolution ("and, logo, original") , The most common are frequencies of about 3.5 to 7.5 MHz, some departments such as dermatology but also use frequencies up to 20 MHz and "exotic"'s also more.

Transducers generate signals using piezo crystals that can transmit / receive the high-frequency sound in an extremely fast alternating manner and are mounted on the "sound surface" at the transducer "in front" - basically in 2D transducers in a series (and in modern) (matrix) Construction types of 3D / 4D transducers in a surface arrangement. The number of crystals significantly defines the achievable geometric resolution (which indicates limitations or difficulties in the development of high-resolution matrix 3D / 4D transducers). Physical (versus "further back" in the signal processing chain or its relevant factors up to the immediately visualizing monitor) are here both the lateral resolution (ie parallel to the sound surface of a linear transducer or corresponding concentric to the sound surface of a convex / curved transducers) as well as the axial resolution (orthogonal to the transducer sound surface or in the propagation direction of the sound waves) have a cardinal quality-defining role, both depending on the distance to the sound surface, divided into 3 categories: from the near field to the optimal focus distance on the scanner adjustable focus zone to the far field, plus the gain as a penetration depth-dependent gain control The acoustic "coupling" to the patient using gel (gelled water). The spatial transducer positioning on the skin is called Anlotung. The transducer sound surface also defines the conversion into the resulting image, the so-called sound field: in the case of 2D scanners, a straight sound surface generates a linear sound field in the form of a rectangle, a curved sound surface a convex (also: curved) sound field, which consequently like the working surface of a single conventional disk Wiper looks - but depending on the 2D application range with coverage angle of 45 ° to 270 ° (Fig. 1).

Correspondingly, 3D / 4D transducers as "3dimensional sound field", more correctly with sound space (later referred to as intensity values here under "IVS", also referred to as "sound value space") - currently typically a pyramid, more precisely a truncated pyramid , or similar shaped volumes (as incidentally for the representation of the fundamental, theoretical principle of operation insignificant, the cut tip (stump) in the following mathematical "IVS" - negligible: The shape of the sound space is practically independent of the respective transducer or The sound field can also already be influenced by the functional principle of the 3D / 4D transducer - there are currently two predominant types:

(1) 3D / 4D transducers, here first the B-mode technology-based volumetric transducers, have in their housing a motor that "swivels" a linear or convex 2D transducer back and forth quickly "Tilting Transducer") (see Fig. 2). The comparison (with reference to a convex transducer component) to a wagging fan: "Fan = original / native 2D sound field, hand = motor" gives a good idea of the functional principle - "aired air area" = "resulting (3dimensional) sound space" - after calculated conversion for the purposes of visualization - ie: Many or all 2D snapshots of the periodic "fan positions" are converted into a three-dimensional image homogeneously filled with voxels, the result volume.

In principle, movement artifacts can also occur in this method, but due to the relatively high engine speed, in practice, there is "little time" for motion artifacts in a result volume - significantly less than for scans with the (manual) "freehand sweep explained later ", which translates into very good results for the" clinical reality ".

The advantage of this method is currently basically a relatively good geometric resolution. Basically, however, the engine and the conversion calculation are both speed-limiting factors for a "live" or "real-time 4D" method for fast moving sound objects. (2) The principle more modern technique are so-called matrix transducers (see Fig. 3): Piezo crystals are attached directly to the desired volume on the sound surface 2dimensionally (just as on the sound surface side by side or one-dimensionally arranged 2-dimensional crystals generate a three-dimensional image) and operate accordingly: each "set of simultaneous sound impulses" generates a native volume without subsequent conversion calculation. where it comes to high image, ie more accurate: volume rates - currently about 20 / sec are reached.The question of suitability for other applications will depend crucially on whether the current geometric resolution for such applications can be sufficiently increased - which is identical for the time being with the question might be, if more crystals can be attached to the sound surface of such a matrix transducer.

Neither is it clear today "which of the two approaches is racing", or whether both remain, nor can further and future 3D processes be excluded - whether mechanically with, for example, a fast rotating "transducing roller", or electronically e.g. with two or more "but" shifted (i.e., space-using) "piezo matrices", - possibly defining additional 3D / 4D transducer categories - or whatever else.

The allusions to sound-conducting properties (see scanner) indicate that there are good and less good (or "impossible") so-called "sound windows", ie for certain visualization purposes sufficiently large and sufficiently far inward sound conducting body regions for placing the transducer - the heart, surrounded in many places by the air-filled lung and ribs, is a "classic" example of the need to sometimes select (few) suitable sound windows, which is a key reason why transducers vary greatly depending on the application, Also in the sound surface (where the piezo crystals are attached) are formed - a transducer, for example, to visualize a large part of the heart through a small sound window between two ribs, is accordingly a small sound surface with a relatively wide angled sound field (= sufficiently large convexity) endovaginal probes are very lan formed with a small, but up to 180 degrees covering convex sound surface; for the closely under the skin lying thyroid (left and right neck area) is suitable a 7.5 MHz linear probe (straight sound surface and low penetration depth at high resolution), etc.

Analogously also briefly the various associated scanning methods, in medical terms: ultrasound modalities (possibly also English: scan modalities) including the term, here chronologically-historical and quasi-coarse also "ascending to technical sophistication" listed:

A-Mode: A single "sonic beam" - today rather history / only little used.

M-mode (M = Motion): Like A-mode, but plotted on a timeline (something similar to ECG, instead of "result curve" but a "result image / diagram), very good time resolution, plays (practically only ) in cardiology / angiology a certain role for the analysis of faster (eg heart valve) movements (sa below: echocardiography, etc.).

A-mode and M-mode may have special transducers. Typically, however, the M-mode is carried out by the B-mode transducer explained below.

By far the most widely used and (also for simulation) most important method is by far the B-mode (B = Brightness): instead of a single sound beam, many such sound beams are emitted / received or respectively piezocrystals are linearly implemented on a "bow" side by side, what can be visualized (due to the multitude) as a two-dimensional sound field - the 2-dimensional "sectional image" very frequently mentioned in the ultrasound, which is why the B-mode is often referred to as "2D".

The B-mode generates the "typical" ultrasound gray-scale image, which is evaluated technically-qualitatively according to geometric, temporal and gray-scale resolution The gray values sometimes also have a (one-) colored (typically golden-brown) intensity scale which, however, has nothing to do with other parametric color images in the ultrasound, which are then also "really multicolored" color codings of their visualized parameters (eg color Doppler, see below). Explanations of the geometric resolutions here relate primarily to the B-mode or 3D / 4D (see below), the geometric resolution of the 2-dimensional (color) Doppler is lower and as a quality criterion of less relevance than in the former method. In cardiology / angiology also the temporal resolution plays a significant role, as already mentioned.

According to the physically given resolution limits, 786 lines were selected in the visualization for the axial resolution on monitors, and then to correspond to this standard in the lateral resolution (horizontal), resolution of two-dimensional digital ultrasound images was 786x536 pixels (monitors with 4: 3 Aspect ratio) - ie just under 1/2 million pixels - which, however, primarily only says something about the monitor display, but is nevertheless tuned to the physical resolution of the ultrasound; however, further physical "subtleties" of ultrasound itself are unlikely to touch on questions of this patenting in the first place Since physics has not changed since then, this standard has at least been maintained unchanged, even though the other image quality has massively improved within the scope of this resolution - also in medicine applies: Geometric resolution is not the sole determinant of image quality.

Doppler uses the well-known speed measurement method, applied to ultrasound: it measures the frequency shift, the speed of a "viewed" object (in medicine or in diagnostic Doppler, these are the blood corpuscles of the flowing blood in sufficiently large vessels After B-mode Doppler is the most important and so far only further method used in the simulation.Doppler is mainly used in cardiology (echocardiography) and angiology (vascular sonography) because angiology tends to be more "free" vessels, less so However, Doppler has also become increasingly important in other ultrasonography (eg, the usually more static liver is perfused in its vessels with a registrable "flow" (blood flow vector), also vitality or therapeutic success A tumor can be diagnosed on the basis of its vascularization, ie blood vessel formation It can only be judged by the Doppler and partly, at least appraisingly, quantified, etc.). Like M-mode, the Doppler has not been explained here in advance for the transducers. This is because this method (such as M-mode) is often carried out by the B-mode transducer as well. (Pure - one-dimensional - Doppler Transducers see, due to only a "sonic beam", so like small tubes off, but find only in the pure vascular diagnosis / angiology a role.) So here's a lot to the procedural basics, but also to the clinical application of the Doppler.

For

/ = Frequency change (Doppler shift)

/ o = emitted frequency (transducer frequency) v = blood flow velocity α = angle between sound direction and blood flow (flow) c = speed of sound in the tissue

the Doppler equation applies: / _* = / <> c

The transmission frequency of the transducer (/ o) is constant here, as much as the

Speed of sound (c) in the body tissue, so that in practice simplified:

/ = 2v -cos «

However, an "axis-appropriate" approach ideally requires an angle of 0 ° or 180 ° between flow and viewing (Doppler transposition / Anlotung) in non-axis-appropriate Anlotung the blood flow (flow), the Dopplershift changes according to the cosine of the Anschαllungswinkels ( α) between sound direction and blood flow.

(As in the standard literature on an indexing of the here actually meant "false angle" in contrast to the above already used or desired "correct" Anschαllungs - angle (α) is omitted, was also omitted here.). Depending on the angle of attack or its cosine, the so-called error width (% flow) is calculated here with respect to the maximum speed to be measured (ie thus the effect of the "false-slang angle"), as in the following table for (in clinical practice meaningful) 15 ° increments is shown: a = 0 °, cosα = 1,00 => 0% Error width a = 15 °, cosα = 0,97 => 3% Error width a = 30 °, cos a = 0,87 => 13% Fe here width a = 45 °, cos a = 0,71 => 29% Error width a = 60 °, cos a = 0, 50 => 50% Error width a = 75 °, cos a = 0,26 => 74% Error width a = 90 ° , cos a = 0.00 => \ 00% error width

From a scientific point of view, this may cause "a bad feeling" for the use of Doppler: It is obvious that in Doppler in medicine too low readings are likely to occur, if not just very accurately sounded. by way of example only, aspects such as "the next emergency has to be dealt with in 5 minutes and the ongoing investigation terminated" implies - actually occurring fairly regularly. Here, however, it should be taken into account that "nominal values" and "actual values" play a significant role in medicine - not only in "scientific papers" (whose data were often collected under "clinical conditions"), but also in form practically or clinically learned "orders of magnitude" with which one has learned to deal (or "should have"). In addition, it should also be obvious that the ultrasound Doppler per se is not an extremely precise technique - apart from some physical-physiological factors just because the first designed for an anatomical / morphological and thus more "qualitative" visualization, manual In this case, analogue (2D) transducers are also used for a more "quantifying" method - which, in terms of the approach (manual transducer guidance), has corresponding limitations for a reproducible-accurate measurement method. (Although you can perform geometric measurements in the B-mode slice images as well, the sono-anatomy / morphology visualization is clearly in first place overall, while Doppler is the first to capture the flow - even if it is a 2-dimensional generation variant Images such as the color Doppler are - in the approach but just "a little more measuring or quantifying" is designed.) In any case - if a method in practical application tends to provide too low readings, so in the clinical use quasi an "offset -Approach ", and the experienced clinician operates markedly with" relations. "If the empirical benchmark tends to be too low, then the individual will If necessary, the measured value is "automatically" calibrated (ie typically also assessing) by the physician. It is almost typical of medicine that such mistakes are "tolerated" and the approach prevails "Such mistakes are not bad as long as they are systematic". In fact, some are very good clinicians, whose pragmatic approach would probably "stun" many a "classical" scientist at first (to say the least), also because excellent physicians, because they have a high level of experience for a reliable-true stratification or Although diagnostics still have a pronounced method-critical awareness, this is unfortunately not the case for all doctors. According to these indications, it should no longer be so surprising that attempts to compensate for the plotting error with a mathematical correction have failed in cardiology for the time being, especially as an effect was then overestimated in an experiment - in which angiology is However, there is now a certain rethinking. For the Doppler, however, it is true that, similar to the B-mode, today's medicine is indispensable, and that this method is clinically very powerful - and in comparison with alternative examinations - despite these limitations nevertheless proved enormously.

In the first place, even in such an introduction, these explanations may seem a bit "ludicrous" or almost derailed on an anecdotal level, but in fact this example has been carried so far because it is truly representative of real or clinical medicine in this capacity, also to an understanding of the wider environment ("this is how real medicine works") should not be underestimated: even with the technical "design" of this simulator, which should finally provide a qualification for practical ultrasound , the consideration of this environment in addition to (hopefully decent) theoretical and technical fundamentals, etc., sometimes plays a decisive role, especially in the realization of the process, further details later eg also under the corresponding HUMR sub-procedure "DAS - Doppler Acquisition and Simulation" ,

Generally still to the understanding: The max. Flow of blood (=> Doppler) in the

Transitional area of the aorta of the heart is more than 10 m / sec., And in particular the heart valve movements can only be achieved from about 100 images / sec. reasonably well resolved (typically => M-mode or -jsphr fast B-mode scanner). For the sen (typically => M-mode or very fast B-mode scanner). For the other, ie heart muscle and blood vessel movement suffice from about 20 frames / sec. (=> B-mode or fast 4D) for diagnostics.

Doppler has a considerable number of variants:

CW Doppler (Continuous Wave Doppler), in which a continuous sound emission / reflection registration of the signals occurs - similar to the M-mode, but on both sides of a zero line generating a diagram image. Thus, low and high speeds can be determined quite accurately, but a statement about the depth of emergence of the signals is not possible statement.

PW-Doppler (Pulsed-Wave-Doppler), where the velocity is registered over a selectable depth range ("sample volume") The transducer does not work continuously (see above CW) but pulsed with alternating send / receive If the frequency of the Doppler shift is above the frequency of the pulse repetition, the shift is no longer displayed correctly (so-called Nyquist effect), but appears shifted in the registration The Nyquist effect is registered on the other side of the zero line, so the Nyquist effect may be compensated for by a zero-line shift.

HPRF-Doppler (High-Pulse-Repitition-Frequency-Doppler), in which an increased pulse repetition rate, the exact measurement of higher velocities is also possible above the Nyquist limit: The establishment of a second depth range in the sound direction axis of the first depth range (here also Messtor or sample volume) results in a frequency spectrum consisting of the Doppler information of both ports. The "optical result" (diagram) shows information recorded on both sides of the zero line.

(The first three procedures listed here are visualized i.w. as separate monitor "windows" on the scanner or not integrated / overlaid / inserted with other simultaneous - tissue morphological - ultrasound images.)

Color Doppler (English: Color Dopplβr) or FKDS Doppler (color-coded Doppler sonography), in which (similar to the B-mode) a 2dimesional designed PW-sound emission / reflection registration takes place: Here is the Flow direction (Fow) color-coded - typically "red = towards the transducer", blue = off- Transducer-away ", and Nyquist proportions" yellow-green ". A great diagnostic strength of the Doppler color lies in its ability to represent it "superimposed" with the B-mode wabei "inserted" is formulated more correctly: Flow logically occurs only in cavities (eg in vessels called lumens, especially in the heart: Cavum), which are practically black in B-mode, so a visibility-obstructing coverage practically do not occur - the color Doppler is in practice also usually so, as so-called duplex mode (ie FKDS combined with B-mode) realized , Color M mode, in which the Doppler shift is color-coded analogously to the color Doppler but as a one-dimensional sound impulse on the time axis such as M-mode

it is typically visualized together with the M-mode chart image. Tissuβ-Dopplβr (tissue Doppler), which does not measure and visualize the movement of the blood flow: technically a low-pass filter is used, whereas in Doppler a high-pass filter sets the desired frequency spectrum, i. the blood flow, filters out). The tissue Doppler is also implemented as a 2-dimensional color Doppler (not combined with B-mode here). Power Doppler in which only the amplitude, i. Signal intensity is registered and visualized (and not the frequency shift). This method is also of particular importance in contrast-enhanced sonography (where the patient receives organ-specific reflection-enhancing drugs).

Harmony imaging, in which only selected harmonic vibrations (in particular 2nd harmonic) of the reflected transducer transmission frequency are registered / visualized. On the one hand artifacts are lower in this method, the resolution is correspondingly good, on the other hand, the lower signal strength does not generally make the method well applicable - hardly surprising, however, that harmonic imaging acquires special significance in the high-contrast contrast ultrasound (see above) Has.

For the time being, in the simulation primarily power Doppler and, within certain limits, the color Doppler will be relevant, but following further Doppler techniques, as far as feasible, may only be a matter of time (and not very long while). Generally at this point the following:

ROI (Region Of linteresf) is the common term for n-dimensional images, which characterizes a diagnostic-relevant n-dimensional area in an image. The common imaging techniques typically also visualize the nearer or even further anatomical or morphological environment of an ROI. Anatomical landmarks are often relatively easy to find and often define clearly and accurately, and help to find ROIs or difficult-to-find images. Structures, but also in general in the morphological (and sometimes functional) understanding of a picture ROIs are to be clearly differentiated as typical clinical terms, ie terms such as "sonic window", "sample volume", etc., and have a fundamentally different meaning Lead structures and their reproducibility also have significant significance in evaluating the performance of an ultrasound simulator alongside factors such as real-time behavior, resolution parameters, and so on.

In addition here also another rounding stratification of the ultrasound with selected, important special procedures:

Incidentally, transcutaneous ultrasound is also counted among the non-invasive procedures - even the gynecologists' trans- or endo-vaginal songraphy, which is to be assessed as semi-invasive. Ultrasound plays a special role in the visual guidance of hollow needle devices coupled with the transducers for the removal of tissue or cerebrospinal fluid samples - the most transcutaneous ultrasound is coupled with an invasive method (examples include the so-called footpad biopsy) Tissue or Amniocentesβ for amniotic fluid removal). In gastro-enterology, so-called endo-sonography coupled with an (optical) endoscope also plays an important role - an invasive procedure.

In cardiology / angiology, the following methods play a significant role: TEE (for trans-esophageal ecocardiography), ie an "invasive" (penetrating into the patient's body) ultrasound procedure in which a very small transducer is housed in a flexible tube you can swallow. In which In contrast, intravascular ultrasound (IVUS = lintra-vascular ultrasound) allows a micro-transducer - introduced by means of a catheter - to emit sufficiently large vessels "from the inside", ie the lumen and vessel wall - similar to TEE the 3D reconstruction of the image data also plays an important role) - also an invasive procedure TEE and 3D / 4D will be explained in more detail below However, in the simulator under discussion, these special methods are not directly or procedurally considered .

3D / 4D systems: The first 3D / 4D systems in the ultrasound were partly hardly noticed, partly euphoric with big expectations than the ultrasound of the future par excellence. With some distance, a reasonable realism, which has meanwhile set in, can be partly predicted: A basic problem is already that the final visualization medium, namely the monitor, is designed two-dimensionally. Another (and even more significant) problem is that 3D / 4D raises the question of how to represent volumes in terms of their transparency properties: unlike in many other areas of life, only in rarer cases is the but rather, as a matter of prosaic terms, into the patient, which is typically "looking through", but obscuring superimpositions are not desirable, as both circumstances very often lead to it for the visualization of suitable "sectional representations" (ie also cutting through the volumes of 3D / 4D systems), which means that one mostly "lands" again in 2D - apart from a few cases, where there are anatomical limitations, in which cuts from 2D transducer positions not transcutaneously (ie resounding through the superficial skin), but through virtually achievable sections a us 3D / 4D volumes (although there "only" transcutaneously generated) can be visualized. This applies especially to 2D cuts, power Doppler and M-mode. So it was not surprising that diagnostic performance comparisons of 2D versus 3D were very sobering, especially since 2D remains superior in geometrical resolution, on the one hand - and that 3D / 4D offers clear advantages for only a few specific tasks, but on the other hand grown methodical experience and continuously improved technology have now become quite significant. So what will remain for the foreseeable future even if any future 3D monitors are implemented in 3D / 4D scanners are 2-dimensional sectional images - and so on. with the qualification necessity for the interpretation of the same - thus thus the (nevertheless not only relevant or realizable) previous domain of the ultrasonic simulation.

In any case, 3D / 4D was and was the breakthrough for their realizability and will retain this status for the foreseeable future - at least as far as this simulation is based on real ultrasound images.

3D / 4D-volume cuts do not yet reach the geometric resolution quality of B-mode (2D) and honestly do with geometrically even lower-resolution monitors - currently around "240x300 pixels" are more likely than optimistic However, since this does not really brag, all manufacturers of 3D / 4D scanners are holding back with statements, but the above estimate was not quite "out of the hand" (eg from experience with the read-in data as compared with the known data and results of the freehand technique) and "with a few pixels tolerance" will accurately reflect the actual current upper limit.

Surprisingly, the first ultrasound 3D procedure was actually a 4D system - and invasively applied to TEE For the "pure" B-mode used at the beginning of TEE, the transducer was gradually pulled upwards and the entire heart sounded from inside the esophagus (esophagus) from the inside under sound ideal conditions without limits, and this method was developed from about the beginning of the 90s onwards, so that the tube is stiffened / compressed with compressed air after swallowing motorized 1-mm layers as a B-mode image step-by-step and ECG and respiratory-triggered data, then converted that data into a dynamic volume, which meant an hour of post-processing - really not a real-time procedure, but the first 3D / 4D ultrasound system was born ler was Fa. TOMTEC.

It was not until the mid-90s that other commercial 3D / 4D add-on systems followed - and for other applications, more precisely: prenatal medicine for the visualization of the

Fetuses during pregnancy, then also internal medicine and cardiology. In this case, the place data processor set of the stepper motor-controlled TEE This is the procedure for "conventional" echocardiography, so-called TTE (for transhurtic echocardiography = sounding through the thorax), 3D sensors for so-called add-on sys- tems - ie separate PC systems connected to the scanner via FrameGrabber - used to generate 3D / 4D volumes for simulation cuts by means of a so-called "Freehand" image collection (by means of a manually or "free" guided transducer, see below "Sweep", see Fig. 4 and 5). This was realized analogously, only without ECG triggering, also for gynecology / prenatal medicine and internal medicine as further applications of transcutaneous ultrasound (ie sounding through the superficial skin).

Figures 6-9 (Fig. 6-9 by courtesy of the company MedCom GmbH) show such a 3D add-on system (also referred to as 3D upgrade systems), which works according to the Freehand method, together with a mobile scanner and linear transducer schematically or in a test setup with a tube object in a water bath (here: model for vessel bifurcation, ie branching of a vessel in 2 branches). Fig.9 also shows on the left 4 time increments of the acquisition movement before conversion to 3D. A sweep was performed with the Linea transducer, the alternative acquisition with a convex transducer (see Fig.9) was performed as a "tilting movement." Rotation is also possible, but in each case a constant progression of the movement direction must be performed - a "reverse" or even subsequent insertion of frames does not allow these methods. A similar system will continue to be used for image / volume acquisition in simulation. For 4D, an image trigger must also be used, as explained in more detail in the later section "Triggering methods for dynamic volumes" (ECG, FET, etc.).

However, soon after such first "add-on systems" 3D / 4D scanners of the ultrasound device manufacturers came onto the market, which quickly supplanted these add-on systems, because the generation and visualization of volumes in the form of such integrated procedures Due to the methodically fundamental proximity of scanners with motor 3D transducers to an initial manual (add-on) procedure, a sweep (initially generally simply as a "swipe movement" of the transducer) was made significantly easier, faster, and free of artefacts on the skin) - and before "rotation" or "tilting movement" over a scan point as the most important detection modality - are the geometric resolution values of this manual Method also approximately on the order of ("integrated") 3D / 4D scanners.

The Freehand method works slowly, but many frames (per unit of length of the sweep - cm or mm - and absolute relative to the total sweep length can be generated during acquisition - as long as patient movements do not streak through the planned result) In a free-hand scan using a 2D transducer, the image detector performs a sweep approximately orthogonal to the longitudinal axis of the sound surface (= cutting plane) "sweeping over the shutter object to be detected as uniformly as possible" (see Fig. 5) The mathematical domain (ie, as in a "Cartesian cube" or a corresponding data table, which then represents equidistantly parallel image planes, one can also say: regularly-homogeneously distributed voxels) is converted into a volume (see Fig. 7) ). The sweep direction should be maintained - e.g. a "backward movement" is not "managed" by this method. (Here's an anticipation: If 3D / 4D transducers are used, then a sweep for the IVS procedure explained later - that is, the acquisition direction on the skin surface - would then be arbitrary.)

There are a few "pitfalls" in the freehand procedure - which are discussed here for two clinically relevant examples:

(1) When acquiring the frames in the freehand method and during the 2D => 3D conversion, "gaps" are typically created, ie for a geometrically sufficiently resolved visualization of the result volume and its sectional representations, too large voxel spacings, which are calculated by means of so-called Interpolation compensated / filled in. For sufficiently small "gaps" this works visually satisfying. For extreme "jumps", as typically occur when scanning very soft tissue - eg scans of the female breast, so-called mamma sonography - but this method is massively overwhelmed and then produces very "absurd volumes" that have bizarre but no clinical value. Furthermore, the Freehand method therefore requires a sufficiently firm tissue background of the scanned skin - otherwise the calculation (eg interpolation) method - see below "simulator" - completely overwhelmed. (This problem would arise analogously when scanning with IVS - there, for example, with "Mosaic / Matching Method "such as cross-correlation and the like., And is, following below or closer, with the HUMV method" soft tissue scanning "practically solved. However, this method is not only well suited to IVS based on 3D / 4D scanners, but also for freehand scans via 2D - especially in motorized variants.

(2) In a 2D transducer, if a sweep direction were chosen "near parallels" to the cutting plane, this would also produce "absurd" results if it were to produce volume. In 2D sweeps for volume generation, therefore, a well-chosen direction (see above) is, in addition to continuous, not too high and not too slow sweep speed, an essential prerequisite for useful results - which is considerable skill of the author (technical, manual, "influencing patient cooperation"). ) requires.

A sweep parallel to the cutting plane, quasi to the (about iateralen) extension of the sectional image (in practice, for example, cut through a leg on both sides of the knee joint or "circumvention of the neck) exists in B-mode, however, as a so-called panoramic process - logically, there are, however no conversion into a volume, because the goal / result is again one (then "very broad")

Sectional view or 2D. This method has some basic analogies to IVS, but was just related to 2D and not as comprehensive or universal or "hybrid" designed, and is often geometrically only approximatively working methodically (realized by a given scan / sweep - Speed - ie not for accurate measurements in the resulting

Sectional view).

Conclusion: The advantage of the Freehand method for volume extraction from 2D slices versus 3D / 4D scanners: Relatively large achievable volumes, disadvantage versus 3D / 4D scanner: significantly higher risk of movement artifacts in the volumes (=> very high workload in the case of detection). Incidentally, this was also the only image acquisition method for simulation at the outset, but because of the large acquisition effort, native 3D / 4D scanner data was also read in, but with the disadvantage of smaller volumes but lower motion artifacts and more uniform image quality - this "dual" simulation technology and quality is the technical status quo of the "first generation" of ultrasonic simulators.

In general here for a stratification of 3D / 4D scanners with regard to their speed:

4D scanners for gynecology / prenatal medicine today have a temporal resolution of 3-5 volumes / sec. (which is sufficient for the medically anyway rather irrelevant movements in a fetus), while the fastest 4D scanners in cardiology currently about 20 volumes / sec. reach ("Realtime 4D") but where fast 2D scanners come to well over 100 frames / sec.

Especially for procedural realizations of the simulation is to understand two things also important:

Also, 3D / 4D scanners offer so far no so-called "streaming" ("live download") of their image / volume data, but only a data transfer to Bildbzw. Volume acquisition. The acquisition of the volume data or portions thereof must therefore be provided with a suitable control, and the acquisition of the associated position data must be coordinated therewith. According to previous practical research experience with 3D / 4D scanners, volume transducers apparently do not form a geometrically "distortion-free" geometry.

However, the extent to which Matix transducers are more suitable in this respect is not yet known, but due to their design, they can hope for "cleaner" results in this respect.

Last, but not least: the geometrical resolution results of all 3D / 4D methods sound quite sobering in a technical sense, but you have to add here in general that apart from the X-ray (where up to 6 million pixels might be useful in large exposures like "thorax") the alternative radiological imaging techniques (such as CT and MR) are not yet "better by dimensions", and the ultrasound - even with this, measured, for example, on modern TFT monitors with 2 million pixels, not to mention digital cameras with up to 10 million pixels, relatively modest resolution - (judged by experts :) can nevertheless achieve a tremendous diagnostic performance, as parallel examinations with alternative radiological methods (also in many studies) repeatedly prove. In addition, there is always the fact that every imaging technique has specific image strengths and weaknesses or diagnostic focuses ("what exactly do you want to visualize?") - these indications are also intended to conclude a weighting "assessment" on the relevance of ultrasound for the time being himself as well as his simulation for medical education and quality assurance / control. Simulator is (here! :) the terminus of a system, in which primarily neither the known methods of virtual reality (simulation based on "computer graphics") or augmented reality (with "real images such as cameras enriched virtual reality") are used , In the absence of a suitable common technical term, I have referred to the applied basic method as "Restitutive Reality".

The fundamental goal of this simulation is

1. a "high-fidelity reproduction" or "restitution" of potentially as much as possible (ultrasound) images - for each individual patient ("case") or

Case record - as close as possible to handling and quality of a real scanner -,

2. not only the imaging process (ie the scanners in the form of a "GUI" or scanner device imitation - including the transducer as the most essential interaction medium in "virtual sounding" in the form of a transducer imitation (also: Peudo transducer or transducer replica), but also the patient (s) in the form of a patient doll (also referred to as a sound model, scan model, phantom or manequin) In any case, as far as sonographic and didactically relevant aspects are concerned, in general such situational and device "imitations" are also referred to as "mock-ups." If one now wanted to include the entire mock-up for a more descriptive term , one could use the term "Augmented Restitutive Reality" analogously eg for the entire methodology - however here with the clear indication that "Augmented" refers to such a mock-up of scanner and sound model.

This simulation, which is used in ultrasound, is methodically and as a method further characterized in that each simulator or each simulator class has at least one recording system or a "recorder" (consisting, see below, scanner, PC, etc.) - he is technically-apparatus-separate or integrated with the actual realized simulator -. The actual simulator can thus essentially be described as a "player", more precisely as an "interactive player". So far as all necessary "Fäl- Ie" / images (volumes) were recorded for the respective training task, the existence of the recording system, however, then - ie for the subsequent simulation - no longer necessary, including recorder / player analogies of consumers Electronics quite similar.

(Rudimentary ultrasound simulators would be conceivable only with simple reading of volumes and so few functions for simulation processing, that the term "recording system" might be a bit high and then for the sake of simplicity the actual simulator would be attributed - this is supposed to Here, however, are neglected because of the resulting modest simulation results, because so no progress i. Compared to previous ultrasound simulators would be achieved, what this is about.)

The recorder thus takes cases via video output of the scanner, which is equipped with a PC including FrameGrabber card, and with a 3D sensor (βdimensional), which is attached to the images on the transducer and the associated simultaneous location coordinates to the recorded pictures. The first such method typically takes (or continues to take) B-mode (2D) images (or 2D dimensional Doppler) which are subsequently converted (corresponding to a Cartesian cube, ie equidistant-parallel) into a volume of homogeneous voxels. Filling were transformed. "Interfaces" in 2D are compensated / filled in. Subsequently, with the actual simulator, this volume can then be "hidden" in a transducer-imitation via the 3D-sensor, which is then hidden in a transducer-imitation The described homogeneous voxel arrangement allows computationally the realization of a table from which the (the pseudo-transducer or its 3D sensor position) corresponding cutting planes or voxel plane with their gray - / Color value and location information can be quasi "read directly" without additional time consuming computational steps required - only in this way can be achieved in the foreseeable future a perceived as "real-time visualization speed." Such volumes are then in a sound model virtually positioned as they would correspond to an anatomically correct insertion now visualized in a replica of a scanner monitor or in its virtual sound field on the simulator. Areas that exceed the recorded volume are "filled in" in the visualized image by speckles whose adjusted gray values are determined from the respective volume.

Later, instead of the (signal) chain: "Scanner-Video => FrameGrabber + 3D-Sensor + Computer" => Conversion to 3D / 4D "then" Direct "3D / 4D-data sets corresponding volume scanner were read in (- praxi is today with both alternatives, so "double-tracked"). Native volumes have hardly any motion artifacts, but the disadvantage that they are limited to the sonic space of a single selected or static transducer position - a "mosaicing" (ie seamless and thereby sufficiently "joining" several volumes), the quasi one Moving transducer guidance (as in 2D, for example, the Panoramic method) is not possible.

Incidentally, the sound models used hitherto in ultrasonic simulation have always been passive, i. lacking any intelligence of their own - and merely offering a patient-like surface to set up the Transducers Imitation.

The statements made so far already suggest here that the simulation - in the strictest sense as cutting control and visualization by a volume describable - technically not so complex and the simulation quality is decisive as the "upstream" recording and recording possibly The following post-processing technique primarily determines the image and simulation quality of the slices on which the slices are based: this note should also be understood as a transition to the next, actual topic here - the technology of a subsequent simulator generation:

By "mosaicing" (eg via spatial coordinates during acquisition, cross-correlation in "matching" and similar methods without gaps, "correct" or "fitting"), volumes are called "meta-volumes" - there can only be one or more per "case record". Instead of magnetic-field-based, other, for example, optical 3D sensors are conceivable - even a sensorless detection of volumes that are larger than the sound space of a volume transducer, by using image processing processors, which "track" the individual voxels (Tracing ) and can connect correctly, would be conceivable. The ability to create or "fake" meta-volumes - for all relevant sound conditions in the most important medical disciplines for visualization and simulation with all their relevant ultrasound modalities (!) - in particular from volume datasets corresponding 3D / 4D scanner - is the subject of this patent application.

(a) A central point is the mosaicing, as HUMR sub-procedure later here called IVS, which in principle generates geometrically correct and arbitrarily mixed static and dynamic volumes integrating meta-volumes. A parallel object is the alternative method DVA, all scanned individual volumes of a "case" / data set without "mosaicing" - or only with limited "mosaicing" or MetaVolumen generation (ie in combination of both approaches) - with sufficient speed ( Real-time effect) can be loaded and visualized from the RAM or other suitable storage medium.These two alternative methods or their combination have not, as far as I am aware, been described or even realized.

Of similar fundamental importance to the overall methodology in this patent application is (b) the inclusion of scanned surface position data of all sound model classes, together with derived position and motion data of the transducer fiducial (as HUMR technique under "STP - Surface-Scan Related and Transducer Processing "and possibly resulting passive, but also reactive and active sound-model surface changes (as HUMR sub-method" AHS - Active and Haptic Scan Models "shown in more detail).

(c) All other methods are, in this aspect, logically-structured following, but by no means less important, but necessary, supplementary subject-specific realization procedures (gynecology and prenatal medicine, cardiology and angiology etc.) to achieve an adequate realism of all relevant ultrasound -Modalities (B-mode, Doppler etc.).

The latter are / are possibly realized modularly in the technical realization, but nevertheless realized completely integrated - in other words: the methods (a) and (b) are applied to the methods under (c) and implemented adapting-integrated.

However, some methods / methods under (c) are also applicable to the "conventional" 2D => 3D / 4D method with manual sweep, possibly also in a motorized version of this otherwise manual method. As is obvious from this introduction, "ultrasound" is a generic term, but only in the narrowest definition of the term is a correspondingly broadly defined uniform methodical and technical-technical procedure based on emitted (ultra-) sound waves and their registered / visualized (interface) reflection, and that a scanner with a transducer attached to the patient is used as the apparatus method - and hardly anything else and their patients, problem definitions and modalities, transducer and scanner types diversified, as previously explained in "more detail" - and this is very different, even more pronounced than for example in X-ray, CT or MR.

Just as there are fundamental similarities and many applicative diversifications in ultrasound, so it is logical for the simulation. Uniform is the mock-up of a scanner, already with sound models (patient dolls) and transducer imitations diversification begins - uniform is the scanning of Schall-Modell-Oberfiächen, but according to their types or classes that are different. The basic principle of ultrasound is uniform, the simulation of an image in the form of volume data acquisition (for the simulation of the B-mode, but Doppler is already a somewhat modified process), further processing such as meta-volume generation and / or volume random access - as well as sectional image visualization controlled by transducer-imitat position.

The further implementation of the method is then again dependent on the respective clinical application or ultrasound modality - and a "good simulator" will inevitably be able to be realized only with a hybrid method, ie considering many relevant application specifics, if "ultrasound "should be at least reasonably meaningful for its simulation - as this patent application reflects. In response to a hypothetical question as to whether a single "stroke of genius" might not be able to solve all simulation problems, the following paragraphs are only implicitly marginal, but they exemplify a fundamental problem. Here are some basic considerations for the procedure subsumed under (a) above:

An ideal simulator requires an ideal acquisition system (logical consequence of the description of a "Restitutive Reality" system), which includes an (ideal) scanner.A scanner ideal for the simulation of data acquisition could simultaneously take about 0.05 sec (Duration for generating a single B-mode image) or (depending on the type) about 0.05 to 0.3 sec. (Duration for generating a single 3-dimensional volume) the whole patient or at least all parts of interest with all the desired modalities ( B-mode, Doppler, etc.) - much as an "X-ray shot" can do.

However, the reality is that a scanner with a transducer that has a sound field or a sound space of limited size and needs to be moved outside of that sound field or sound space for visualization, per se, can not have this capability - of different modalities to talk. For the diagnostics, this is not needed either - you just move the transducer to where you want to see something - you could also say, with every transducer movement in the "normal diagnostic scanning" you set the clock back to zero, so to speak. for the simulation) there will be no "ideal scanner" for the foreseeable future. This requirement for a scanner thus arises only through the desire to use image data for the simulation (apart from the "panorama mode", which does not meet the simulation requirements adequately.) However, the elapsed time during the acquisition generates - whether 2D - or 3D scanners - in the following simulation artifacts (for 2D => 3D / 4D within a volume, and for native 3D / 4D data when "mosaicing" multiple volumes as a "mismatch"), "because the patient is alive".

However, images of an absolutely sedated patient can not provide an imaging procedure, as long as the patient is alive - with slight motion blur, for example, X-ray, CT or MR must live the same way - it's more or less the same extent. In this case, an attempt is made with HUMR to achieve approximately the same level for images / volumes that exceed the size of a transducer sound field or sound space, such a "level" of "blurring freedom" overall, which corresponds to other methods or below one perceived as disturbing Boundary lies (and otherwise to ensure a sufficient realism in simulation of an ultrasound examination).

In this sense, the patient can not keep his whole body sufficiently quiet for a long while, and even the "living body" involves uncontrollable small, but often artifact-generating movements - such as in 3D / 4D data acquisition For example, "panorama" images, which have virtually no such perceivable artifacts suggest, but apparently sufficiently "suitable" reconstructed for meta-volumes, which has a direct relationship with the size of 3D / 4D volume: The "Panorama Technique" works only in 2D, but i. Compare to 3D processing methods with a fundamental difference, which gives the decisive hint here (also in the practical results): Here, no sweep is made approximately orthogonal to the sound field surface, but i. See with sweeps to generate volume just "approx. 90 ° twisted ", for example (with ideal capture exactly :) in the lateral axis of the 2D image - more prosaic: The 2D images are not" one behind the other ", but" side by side "put together - and the result images are downright" captivatingly clean "(artifact-free) and high-resolution.

For a comparative analysis, the simplification is assumed that the extent, also with respect to "maxima", of artefact-inducing movements is approximately the same - but this is not unrealistic or roughly correlated with practical experiences For example, it may be assumed that such artifact-producing motions have approximately a "back-and-forth" motion pattern, which is critically important: such movements (whether small twitches or the circulation or pulsations of the bloodstream) are not approximately linearly progressive movements but return to their point of departure-typically and approximately descriptively-they are typically (1) more or less fast "swinging" movements around the ideal resting state, which in practice also have upper amplitude limits when no extraordinary events occur.

Although this is somewhat simplistic, it is nevertheless fundamentally important for understanding computational motion corrections. Somewhat more prosaic: In the normal case or in the statistical normal distribution (typically) you do not have to calculate a 10-fold greater correction after 5 seconds than after 0.5 seconds, just a "something others ", possibly also" very similar ", and in practice within absolutely compensatable" amplitude "or" vector "limits of such movements (any comments on non-linear equations describing" more complex "movements certainly do not belong in this context Therefore, a longer recording duration does not necessarily imply per se larger corrections of movement - at least to a certain extent one can say: the larger the image / volume components, the better the result, because fewer transitions have to be calculated (between which the native image / volume is maintained in its quality), or essentially the principle applies: the less such corrections are necessary, the better the result.

Again, a simplification that certainly only applies to a certain optimum and also leads to worse results.

For example, assume the width (lateral axis) of a "real" (ie not the monitor image) 2D sound field with 30mm - and consider for a comparison now the distances of "consecutively" arranged frames (2D) in a manual / "conventionally" performed sweep (approximately orthogonal to the sound field lateral axis) for the subsequent conversion to a volume - assuming (realistically) that at 30mm "sweep distance" 90 frames were captured (corresponding to "frame"). Rate "= 30 / sec ^* " Sweep Speed "= 10mm / sec. ^* " Sweep Duration "= 3 seα), the following becomes obvious:

In the case of frames (2D), for the purpose of calculating volumes (with respect to the desired arrangement as in a Cartesian cube), "irregular" spatial image arrangements must be compensated for - by equidistant and parallelism-producing and gap-filling interpolation algorithms The same assumption for at least 90 frames - and every transition from one frame to the next generates "blurring" or image blurring - thus, in a conversion, 2D => 3D over 30mm "sweep-distance" will theoretically also be 90 times "blurring" or image blurring "into the volume".

If one compares the methodology for "panorama images" with their "i.V. 90 ° rotated or parallel / along the image lateral axis carried out sweep" on, it can be seen that in the optimal case (ie in the case of even skin surface or linear sweep), the next "insertion" must be made only after 30 mm.For the detection of even sound structures under non-flat skin surfaces in practice also significantly more (depending on the process, possibly partial) frames are joined together also significantly more (depending on the process, possibly partial) frames have to be joined, but still far less than the "frame-to-frame" distance in a sweep for conversion into a volume requires.The crucial point here is that the panoramic The process produces much "sharper" images, which also has its cause in that much fewer "transitions" must be calculated by image technology - even if one does not in practice on the theoretical value of "90 times less transition calculations" according to the example Assumptions come - "zigly" less transitional calculations come out nevertheless, and as well as the theoretically employed consideration here also the practical results look.

Again, the initial assumption of the typical characteristic of artifact-producing motions is explicitly pointed out, and then it becomes obvious: While the Panorama method benefits from this realistic assumption in the form of excellent (ie practically artifact-free and high-resolution) image results, this can be said of not say conventional 2D => 3D conversion.

Rather prosaically-critical and expressed somewhat ironically, the last-mentioned method, with great diligence, calculates blurring "mm by mm" into the resulting volumes, which is no longer necessary in the current state of scanner technology, and it does not take advantage of the fact that transitions Even after much larger "sweep distances" would be possible, sufficient and no more difficult than transitions of small distances, because "consecutively arranged" frames do not allow this, although it could massively improve the resulting voxel resolution thereby also the disadvantage of high risk 3D / 4D scanners deliver much more artifact-free volumes within the transducer's sonic space - for use in simulation, such volumes are either gapless and virtually imperceptible, similar to the panoramic technique Transitions "side to converge "or to achieve such" mosaicing "in the visual effect by rapid direct access to all volumes necessary for a complete data set, but actually make it obsolete - or to realize a combined process of these two approaches according to the results of appropriate research preparatory work for it ,

"Mosaicing" volumes may be a bit more complex than frames (2D), but there is no plausible reason to assume that this is an insurmountable difficulty - quite the opposite, on the contrary: the panorama method yields one important reason to assume that the methods of this simulation application will achieve a similar quality leap in meta-volume computation, as already achieved by reading in native 3D / 4D datasets, but limited to sonic space so far a static transducer position, ie associated with the purchase of a significant disadvantage. The procedures of the proposal suggest an achievable size for meta-volumes that is significantly beyond the level of previous conventional sweeps for the 2D => 3D / 4D method. Rather, this theoretical consideration, like the practical example of the Panorama method, suggests that limiting factors will ultimately be the lower geometric resolution or other weaknesses of 3D / 4D scanners (compared to 2D scanners) Solving the problem of volume transitions - after all, the simulation of ultrasound would, technologically, have virtually re-established the connection to the real current scanner technology, and these have been perfected ever since decades.

Finally, it should also be pointed out that, as a result of the considerably lower susceptibility to movement artefacts, the registration according to the methods of this application also becomes significantly simpler and "less stressful" for the acquiring physician - and likewise that this obviously constitutes a "subsequent or subsequent improvement" Insertion "possibly partially" stunted "meta-volumes (ie proportionate volumes) should be feasible - these are rather practical" side effects "whose real importance in the short-lived clinical reality but are extremely high: Without recordings is also the best technical simulator worthless.

Finally, for the clinical and didactic sense of a visualization of meta-volumes - they are calculated as "post-processing" after acquisition or before the actual simulation or realized as an effect of a real-time visualizing memory access to all relevant individual volumes or "combined" : The previous simulation was focused on "post-graduate education", ie trainees who typically already have a lot of sono-anatomical (pre-) knowledge, although there was often the desire for volumes, which is often dependent on the case Often, cases could not even be captured as completely as they should be in relation to the ROI, and anatomical lead structures, for example, can often only be visualized within the ROI as far as they are available. that the previous alternative of "direct" reading native 3D / 4D Data sets from volume scanners do not solve this problem, but often even lag behind in this respect, even though these small volumes hardly contain movement artifacts. The more the simulation applies to "beginners", the more serious becomes the limitation of the visualizable volume size, but the better incorporation of the basic training is an important objective of the simulation, even for a later inclusion of comparison methods, eg simultaneous sections of the same case by a " Quasi-parallel recorded CT volume as well as simultaneous sections through anatomy-didactic, eg graphical or histological based models and so on, can only be practically realized with meta-volumes. There is nothing more to be said here - the clinical and didactic benefit is certainly very significant and virtually indisputable - apart from sometimes somewhat "perfectionist" aspects such as the demands of some technological or medical experts to be able to simulate a scanner and patient as realistically as possible in general. At any rate, it is clear that the corresponding demands will certainly increase all the more, the more and the longer the simulation is used - that too is ample experience-based - and must be considered "with moderation".

However, the following should be explicitly pointed out: In principle, the use of 3D / 4D scanners is the means of choice for improving the simulation. However, HUMR is not entirely focused on 3D / 4D scanning data: both. Mammary sonography as well as (pediatric / adult or fetal) cardiology is characterized by the fact that further anatomical environmental representations are of little interest here - and here e.g. The conventional Freehand method can still be useful - not least in motorized variants, which then make the problem of motion artifacts obsolete.

Preliminary remarks on the procedures subsumed under (b) can be dealt with much more quickly - on the one hand because the topic is less complex, but also because there are hardly any alternatives to be weighed here, but almost only filled a previously existing large gap in the simulation becomes.

It depends on the transducer position - ie, the "gross" localization on the patient, to the "fine-tuned" exploration of the ROI and its details - which is taken up (meta) volume for sectional visualization. The reality today is that simulators, so to speak, do not know their "patients," or more precisely, the surface of their patient models ("dolls"). In other words: the Software knows only one volume, which is manually positioned so that it is anatomically correctly positioned in the sound model, and the relative position of the transducer. From a consideration of physiologically given "sound windows", "perspective" differing different transducer positions, possibly resulting passive, but also reactive and active patient movements and the like, there can be no question in today's simulation - the software knows nothing about the position of the transducer on the patient, its sonic nature - and certainly not of organ displacements, as they are not simply "present" in abdominal sound, for example, but quite specifically for a selective visualization by the impressions of the transducer of the resounding Doctor can be specifically used - so resulting organ shifts as well as related patient (re-) actions.

This part of the simulation used to be "rudimentary" even with the most friendly consideration - a volume that was anatomically correct in itself but reasonably suited to the complex reality of ultrasound only under certain conditions (in particular to the transducer positions) had to be used in all situations virtual sounds and training for the reality "serve". Also, the later described "in more detail" method of scanning sonic model classes and so together with the already available position data of the 3D sensor then possible sound model-related derived transducer movement will cause a significant jump in quality in the simulation The deficiencies described above can thus be significantly improved or even eliminated.This is not listed here and repeated - so is enough for an introduction enough.

The methods subsumed under (c) above are then explained directly in their individual representations so as not to go beyond a reasonable scope of an introduction. As will become even clearer in the following - non-technical - final chapter, ultrasound simulation is a process that has developed a momentum that has become almost unstoppable and that will tend to intensify in the future. It is obvious that in the age of the Internet, a broadening of this process will be imposed in two ways: Not only will more or less singular centers per department want to record cases and make them available for training, many centers will want to participate.

The training will reach a much larger user or "trainee" circle - from students to the "pioneer times" of this simulation.

This will reach a point that does not only define a need for some privileged trainers or training institutions, which will generate both a need for much larger volumes and the need for low-cost simulator systems that will require a corresponding number of users without the can afford today high initial investment.

In the last chapter of HUMR, a solution is shown that corresponds to this expected requirement profile.

Finally, a short, non-technical summary of the relevance and impact of the ultrasound simulator in medicine or clinical reality:

Ultrasound has not yet been integrated into basic medical education - neither in Germany nor, for example, in Germany. in the USA. So far, training is actually a "further education" - or internationally formulated: "post-graduate education" - depending on the respective countries: However, this can also be the training of "Nurses" (analogous to "MTA") or, as in the In Germany, the ultrasound is primarily in the hands of the various specialists - unlike with X-ray or other imaging techniques, the radiologists have no monopoly here - but this is country-specific completely individually regulated, eg in the USA, where it is assigned primarily to the radiologist, their lobby has been able to defend their primacy so far largely for the question "who should and who may sound?" is a "good answer" to the question "Who can provide a qualified education?" of great "strategic" importance:

Medically speaking, there is a great deal of evidence that specialists have their own, immediate (first) results providing imaging, and the ultrasound is with its harmlessness and mobility for this assignment additionally predestined - but here it is also about money or the question "who may use ultrasound diagnostics (eg with - even to the question of how far an "initial" ultrasound examination after admission of a patient or a whole examination procedure could subsequently make further radiological examinations obsolete - thus also for medical (he) -intem professional policy. Training, quality assurance and reproducibility are among the most important arguments of radiologists against ultrasound practiced by other doctors. All this concerns the simulator, in particular, because it is the only method that can be used in practical ultrasound at all a canon of defined cases "What should and what must be (especially in pathologies) seen or can diagnose qualified?""Realpatients" as training objects are increasingly rejected for ethical reasons, except that for many practical reasons this has never "worked" and never functioned so systematically This is already quite a "heavy caliber" in these discussions - the radiologists themselves (except for the USA for their Ultrasound Technicians) do not have any simulator training themselves yet and thus get their own reasoning in now the defensive. Equally plausible, it is certainly not just enthusiasm for this method, but also a lot of resistance - from the "scrutiny" of many a deafening doctor - to the loss of meaning and revenues of ultrasound trainers without a simulator - to fears Immediately such occupational disputes are arguably no argument in a weighting of the meaning of this simulator and possibly therefore a weighting of the meaning of the patent application to be submitted here - but perhaps because of nevertheless achieved successes perhaps nevertheless:

The fact that in less than 3 years now more than 1,000 courses by the German gynecologists have been carried out with this system as one of the largest such quality assurance programs in the medical profession - in Switzerland this is similar - is certainly a proof of this explanation , Incidentally, this only applies to the "core training", plus a similarly large number of courses for more specialized topics outside this program. - Please note: number of courses offered by the FBA (gynecological federal academy), not just participating physicians (maximum 16 participants per course with 4 simulators - very often booked out.) The fact that the largest professional liability insurance with the gynecologists, the Assekuranz AG, now offers a "discount" for the course participants, has no sentimental, but rather actuarial reasons - this quality assurance program has clearly shown the number of claims due to misdiagnosis. (Incidentally, these courses were (logically) done with the "first generation" of simulators.)

These courses were again available for studies and publications on their results - with case numbers of up to 1,000 physicians. In short, the results showed a good acceptance of the simulator and the desire for its further use. With the simulator "in the back", its pioneers ventured then also on the tricky topic, which efficiency conventional courses with "frontal lessons" for the real pathology recognition have - result: An improvement remained in the range of the statistical noise or was not detectable. On the other hand, with the simulator, depending on the subject, quotas of between 25% and 70% were found already after a course for the improvement of pathology recognition - results which correlate with the reaction of said liability insurer.

Since even learned knowledge and skills have no unlimited expiration date, a point will certainly be reached where (in medicine or in ultrasound also usual) "refresher courses" become relevant - apart from "growing generations of doctors" and system- "Updates" with new cases, improved / new ultrasound technologies, which in addition to improved simulation technology (all cases must then be recaptured) will compensate for any "saturation" that may have to be considered. In this respect, these aspects therefore have a lasting character. In addition, the University of Heidelberg is currently starting, for example, a program for the integration of ultrasound with emphasis on the use of the simulator already during the medical study program.

At the last four annual congresses of the German-speaking ultrasound companies (jointly organized by Germany, Austria, Switzerland) DE-GUM-ÖGUM-SGUM, simulator-based training, especially in gastro-enterology and emergency medicine, has already become practical Similar to the DGGG congresses of the gynecologists - as well as to other specialist medical congresses - besides the already mentioned special simulator cousin programs. Also corresponding "awards" and the like - directly for the simulator - or indirectly through the theme of corresponding authors on the simulator - clarify relevance and success:

- Twice FBA training award for the authors or the initiator of the first real-time simulator in the ultrasound (realized by SONOFIT) and "Pioneer" of the simulator in gynecology and prenatal medicine, Prof. Dr. C. Sohn and the " Realizer "of the case content and didactic ob / gyn concept, PD Dr. med. A. Scharf - both at that time MHH Hannover, meanwhile both University of Heidelberg - Habilitation of PD Dr. med. A. Scharf on the topic of ultrasonic simulators in quality assurance and NT diagnostics

The highly acclaimed Merckle Recordati Research Award for Gastroenterology Authors, Dr. Ing. C. Terkamp - employee of Prof. Dr. med. M. J. Gebel, Gastroenterology / Hepatology / Endocrinology of MHH in Hanover, formerly President of DEGUM, the "pioneer" of the simulator in its interpretation for the interior and

Emergency Medicine

Permanent exhibit in the "Cybernarium" Darmstadt, where a cross-section of current, interesting virtual technologies is made available to the public - permanent exhibit in the special exhibition "Computer.Medicine" of the Heinz-Nixdorf-MuseumsForums / HNF to the 100 most interesting computer applications in the Medicine, first Paderborn, then as an international traveling exhibition.

All in all, although positioned in a very inhomogeneous scenario, the simulator has developed sufficient momentum for its future "bottom line", and in addition to a considerable number of publications and serious validation studies, there are also (also in "top ranking" Journals published) a considerable number of diverse "hard facts" on the relevance of this method and the practical initial proof of the system implementing it - even in its "first generation" despite some weaknesses. Apart from SONOFIT, there is only one other manufacturer of ultrasonic simulators worth mentioning - the company MedSim in Florida, USA, which did its development work in Israel and presented the first ultrasound simulator ever in the mid-90s - although this was not realistic. Time simulator. The first real-time simulator was introduced in 2000 by the German company SONOFIT. Even otherwise, the concepts of the two companies are quite different - mainly due to the diverging training focus "Ultrasound Technicians" according to the US "ultrasound culture" - and the training focus "Specialists" according to the Central European "Ultrasound -Culture". However, apart from their "home base", only SONOFIT - largely because of the real-time behavior - has now also found a certain, if not yet resounding, international distribution.Thus, SONOFIT now has installations in Western and Eastern Europe, the USA etc. on all continents except Australia and Africa.

However, one thing must be said: If ultrasound is understood as what it really is when it comes to expert opinion, both companies are currently impatient with the claim of an "ultrasound simulator." Not even the largest application "gynecology and prenatal medicine "is completely pictured: gynecologically, both lack the image of important mamma sonography in a form that also allows training for systematic structured basic examination of the female breast - and prenatal- medically lacking both fetal echocardiography as the most common malformation. Finally, because both of them lack the "heavyweight echocardiography" altogether, one should not even speak of two alternative "sonography simulators" at best - just to clarify the most obvious "absence highlights" - a real "ultrasound" Simulator "should at least cover the most clinically and quantitatively important areas of medicine - corresponding to the real ultrasound - in order to really enable" ultrasound training. "By the way, this would also make a commercially critical mass through a continuous system usage /

Utilization reaches a point where a simulator for many larger hospitals, where these departments are often fully represented (and so would be used by many doctors or "technicians" for training), becomes interesting - and not just for specialty centers and specialist training organizations - it starts consequently, only hesitantly. What is not, but can still be - an exciting question is now just who this essentially wins the race. If such an ultrasound simulator can be realized and then "brought to life" through good and many cases, this will have significant long-term effects on ultrasound and, in addition, some structuring of medicine and healthcare costs. In any case, the further breakthrough of the simulation as a training method will be decisively influenced by its sufficient perfection and completion.Finally, that alone the sono-anatomical imagination as well as the complex interaction "eye-hand-brain" are in need of training, also pathology- Oriented training is a "must", and the conclusion "for the cheapest imaging process even the simplest training enough" is fatal, but now understand more and more responsible in health care.

Overall, there is now a virtually unstoppable increase in acceptance. However, this will (as stated before) also generate the need for new technical concepts including the content, the distribution of case histories (case database), etc., where the simulator moves away from a somewhat elitist instrument to a method and will develop "for everyone" (in the context of medical education / training) a system down - this should then, however, "quality assurance of this quality assurance method simulation ^'in itself" also accompanied by flanking measures are a.

1 HUMR

The first approach to simulation in ultrasound was (and continues to be) characterized on the one hand by overall good acceptance and remarkably good results in training efficiency, also according to many validation studies, and has now become demonstrably on the rise as a training method. However, today's procedure also has a number of disturbing limitations that slow down further success:

The cut sheets have a geometric resolution, which is due to the process in the X- and Y-axis reduced to less than half compared to the real B-mode of 2D scanners - or to 20-30% of the pixels, as far as they "Freehand method" can be captured as 2D images, then converted to 3D and used in simula- tion can then be visualized as a 2D cross-section through the volume. In addition, motion artifacts, often too small volume size and others, interfere with perfection in the simulation - and finally, important modalities in the simulation are lacking - or are so limited that the "ultrasound" could not really be simulated, at least in all important forms.

Although the alternative, a direct digital data acquisition of native 3D / 4D volumes from 3D / 4D scanners, does not provide much better resolution, these volumes often appear significantly "clearer" than resolution would suggest because they are less Movement artifacts have (due to the process due to higher scanning speed) and here correspondingly smaller correction calculations have to be performed, as the "Freehand method implies - the disadvantage of this way, however, is that the volumes thus recorded on the size of a single or are limited in static transducer position: The size of the available volumes for simulation therefore corresponds only to the sound space of the transducer.

As scanner technology has been improving very dynamically for decades, however, further enhancements to the technology such as 3D / 4D scanners have been made. in the geometric resolution, while the "Freehand" method is practically no longer being developed, so it is very likely that now is a good time for a procedural change in the simulation, the quasi on 3D / 4D scanner By the way, superior results are typically expected even today, and HMSR offers solution approaches with the IVS and DVA sub-methods and methods presented here - and for the first time also takes into account other major weaknesses in the field Previous methods in ultrasonic simulation.

Absolute perfection is currently hardly achievable - and the introduction has already been made - but the goal of a simulation, which now sufficiently conveys the impression of real ultrasound and meets all the requirements of ultrasound training by means of simulators, should and can be expected HUMR be achieved. (Other didactic system components should not be treated here, since they are not or may not be the subject of a patent application.) However, HUMR also reflects the experience gained during more than 1,000 courses and their evaluation, and further studies / assessments by Leading ultrasound experts who are recognized as "top addresses" in Germany, Switzerland, Great Britain and the US HUMR thus contains components that are simply the result of the desire for optimized simulation with many years of medical and simulation know-how, and partly Components that have actually grown up in this context - that is, the realization of summarized requirements that have resulted from clinical-didactically identified deficits as well as technical deficits of the previous ultrasound simulation.The relevance of such a target achievement was briefly outlined in the introduction The term HUMR stands for "Hybrid Ultrasound Meta-Volume Restitution" - the last individual terms have already been explained in the introduction. As a general definition, HUMR is writable as follows:

HUMR is a multi-modal method for detecting and restoring largely artifact-free, resolution-preserving, and relevant physiological factors of consider- able static and dynamic ultrasound image volumes ^* - which may exceed the size of the sound field / acoustic space of a transducer ^** their use in visualization and simulation.

* (The order of the named features is arbitrary and no significance order) ** ("Transducer sound field / sound space" refers to a static scan position)

Overall, HUMR is a heuristic or hybrid method, which partly consists of (1) single methods / procedures common to all applications, and partly of (2) individual methods / procedures which only relate to specific applications. As a rule HUMR assumes a (image) volume acquisition using 3D / 4D scanners, eg DVT and SMA can still be used additionally with the Freehand process (2D image acquisition and conversion to 3D / 4D) become. The reasons for this modularity have already been presented in the introduction. Nevertheless, a HUMR simulator is to be understood comprehensively only in its entirety of all methods, and the individual methods will nevertheless also represent a whole in their realization. However, HUMR is - as far as it relates explicitly to SDMD ultrasound scanners, not only dependent on their development, ie primarily their (image) volume quality, but in the process realization of the currently available computer ( PC) technology. Compared with today's simulator generation, the following picture emerges:

The higher voxel number of volumes or slice pixels, which is due to increasing geometric resolution compared to the 2D-based Freehand method with 3D reconstruction versus better and better 3D / 4D scanner, and in particular by the significantly increasing volume size, the This will result in higher computing power and significantly more memory. Parallel processing of additional data, such as the sonic model surface and related movements of the transducer fiduciary, will most likely result in a multi-threading process. Require processing.

However, the time for HUMR or a patent application is deliberately chosen not only with regard to the development of the ultrasonic scanner technology (=> acquisition aspect), but also the computer (PC) technology (=> simulation aspect): we ste - In PC technology now also broad front at the beginning of a change from 32bit to 64bit systems on the one hand, and to multi-core computers (multi-processor systems) on the other - and also the rest is simultaneously the whole PC - Technology, eg Incidentally, this will, in turn, affect the (also PC-based) scanner technology.

So until HUMR has evolved into a production-ready product, from (application) research, to system development, to a sufficient canon of recorded cases, even the technologies necessary for good implementation will all be available at a reasonable price and high performance. Level is available - and the status quo is technologically sufficient to start all this work now: There are already the first 64bit PCs, duo-core processors (for Quatro and Octo versions is already under development), powerful graphics cards , fast RAMs in Gbyte chip size, plates that are faster in throughput than video streaming - as well as SDMD ultrasound scanners. With some HUMR methods, it is actually difficult to predict exactly how the final variants will look in the realization, and "research" is not exaggerated - after all, there is a lot of new territory to be explored: One may have to perform test setups for peripheral components or use "Rapid Prototyping "o. Like. Different algorithms test to get a clear idea of the optimal implementation, if one no longer or less immediately reaches a satisfactory for all essential working conditions solution.

Thus, e.g. DVT (in the form of probably the most difficult sub-method FET), and AHS (for the pneumatics, microprocessor control and a first model of the acoustic model) and SMA (for the most important part, the mammography sonography) already realized as prototypes, that the basic functionality can already be proven, but some final, optimizing work has to be carried out - so research work has already been largely completed. This is not yet the case for ADP, PVP and STP (although these are the ones where the risk of "bad surprises" seems the least) - IVS and DVA have finally begun their preliminary work, which shows the first satisfactory results Results, but there is still a lot to do here.

The alternative methods IVS and DVA are a good example of the procedure, which also results from the amalgamation with the respective, evolving computer technology. DVA is almost the "straight-forward" approach - you simply save all recorded volumes and access them directly in the simulation again.Firstly, IVS is more of a reflection and perfecting of the freehand method with 2D capture to processing native 3D / 4D volumes, both depend on the quality of their implementation on the performance of the usable computer components - which in turn directly affects the algorithms used.

The PVP method is closely related to DVA: Actually, the only difference in the basic idea is that PVP is designed more for meta-volumes - and otherwise that, accordingly, the different transducer perspectives diverge more and thus distances between different to be loaded (Meta It may well not be sufficiently predictable, if IVS or DVA proves to be better in practice - or a combination of two or more volumes. the. The latter seems to be the most likely outcome from today's point of view, but for all the reasons explained, this is rather speculative: the realization of the method should be highly dependent on the available computer technology and the outcome of appropriate research. Therefore, IVS and DVA methods are presented as separate alternatives - with the remark that only one of the two or even a synthesis of both approaches leads to an optimal simulation realization.

The PVP process is presented separately, (1) as it is currently not foreseeable how exactly it will be used: If IVS proved to be the superior variant to DVA, it should in any case - and possibly as an independent software component - used become. If DVA proves to be superior, it would probably be treated as a subset or as a special case of DVA in programming - but it may also be that, similar to IVS, it controls the appropriate portion of the available volumes of a DVA. " Meta-dataset "defines and controls their loading - (2) because the anatomical-sound-technical background of the procedure differs in part: PVP is initially a problem-solving for the different mappings of the same volume from different approaches, whereas DVA in the approach A solution for blasting ultrasound images that exceeds the size of the sonic space of a single volume, however, both approaches are largely congruent as a method solution.

A prerequisite for the proper functioning of IVS, DVA and PVP is the SSP + VTP procedure. The fact that the software not only "knows" the volumes but also each sound model and "understands" the related transducer location data and movements in a suitable form - and this is exactly what SSP + VTP does - will produce a truly satisfying response Simulator regarding the (meta-) volumes to be loaded and the resulting virtual cross-sectional images reachable. SPP and in particular VTP should, as far as necessary and possible, also include a "look-ahead" logic for movements of the Trandsducer-Imitats, all of which could already be a (anyway permanently running) process so computationally intensive own software thread or its own processor core is required if the visualization of virtual cuts should not lose their real-time character - and that requires more computing power as before. On the other hand, with a look-ahead logic via its own thread and processor core, STP could even accelerate the loading of the images and, if necessary, also control other utilities that ensure as "soft transitions" as possible between different volumes. Nevertheless, STP is presented as a stand-alone process, despite the fact that it also has other capabilities that are more extensive than those discussed in more detail in the STP chapter.

DVT treats triggering in the acquisition of dynamic or 4D volumes - especially the sub-procedure of fetal echocardiography triggering (FET). ECG and respiratory triggers, as well as a trigger for irregular movements are briefly mentioned here - e.g. ECG or respiration triggers are not innovations and are therefore neglected.

AHS also deals with the topic of "active" sound models - ie with their own (computer) intelligence - in ultrasound simulation, as well as better visualization of living patients by means of respiration simulation, where this is required for the application (eg Abdominal sonography) as well as "reactive" (eg pain reactions in emergency medicine) and interactive (ie as program control for the simulator) components. In turn, AHS has connections with STP.

ADP is a process related to PVP, but advanced for Doppler sonography. It further distinguishes itself by a dual solution for recording and simulation, which would make no sense in PVP. In principle, ADP covers all Doppler methods in which the angle of attack is affected by the known error angle problem.

Finally, SIA is a procedure that makes the detection of soft tissue without supporting structures (eg the female breast for the clinically highly relevant mammography sonography, possibly also fat abdomen, etc.) possible in the first place meaningful for the simulation. It is basically possible in two variants - one with manual transducer guidance and one with motorized transducer guidance - only in the latter case, a simple control software may be useful. The procedure is otherwise a purely "mechanical" solution, only with SIA is a complete chest (or the two of them), which is actually essential for the basic education in mammography (systematic or structured examination of gland tissue etc.) - In addition, SIA also provides basic prerequisites for a standardized examination - and the absence of a standard has been the main criticism so far at the Moms sonography.

The said "first generation" of ultrasound simulators has in common that at least in its beginning it was based on the conversion of acquired (2D) ultrasound frames into a volume, but on the other hand, that they are based on magnetic-field-based 3D In this representation of a "successor generation", which is based on the HUMR method (s), the latter will initially continue to be based on the latter 3D sensors. Meanwhile, other sensors - e.g. optical, camera or laser-based systems, inertial-based sensors, etc. - conceivable with which the necessary 3D data for acquisition and visualization or simulation including the subsequent positioning of (meta) volumes can be determined. Without spatial data sensors in the detection (but not in the simulation) would also come from processes that use image processing processors that "track" all voxels (and thus volumes (over the size of the volume, by the immediate sound space of a Transducers is defined, beyond) - this would suffice for the pure visualization of large volumes, but this alone is no solution for the simulation of volumes in a sound model.

Alternatives to magnetic field-based 3D sensors are discussed in more detail in the last chapter 10 - but the other HUMR methods are also applicable to them, so that in Chapters 1-9, for simplicity, only the previously used magnetic field-based sensors will be discussed.

HUMR thus serves a purpose perfecting visualization and simulation of imaging ultrasound-fulfilling procedures in many important aspects.

2 IVS - DVA: 2 presentation approaches for meta-volumes

IVS (Integrative Volume Sampling) and DVA (Diract Volume Access) are two alternative methods of virtual simulation of large enough areas of the patient model, if possible. Both are initially alternative methods in which native 3D / 4D volumes are read by appropriate ultrasound scanners into an acquisition system.

IVS and DVA may also gain some importance for diagnostics. Instead of controlling with the 3D sensor (in the transducer imitation), e.g. Also, a mouse can be realized for cutting control through a (meta) volume - even without a patient model. One could scan larger, coherent areas of a patient and also visualize them after the examination or without the presence of the patient - similar to radiological procedures (X-ray, CT, MR) - as far as the patient presence is not for one patient Interaction / cooperation (eg defense tension of the abdominal muscles during ultrasound examination of a biliary colic) between doctor and patient is necessary. An Internet or (W) LAN variant could also be realized on this basis.

For IVS and DVA the following common aspects have to be defined:

Both methods capture 3D / 4D volumes with the goal of having larger contiguous areas immediately available for visualization or for virtual sounding, as this corresponds to the given size of a transducer sound space, for example in a static positioning, even larger than this was possible with the previous Freehand method (2D => 3D / 4D). This would be much closer to the real sound of a patient than the previous simulation. The geometric resolution possible with 3D / 4D transducers should be preserved practically 1: 1. IVS pursues the approach of actually generating such meta-volumes in such a coherent way, while AVS pursues the approach of simulating such an effect through i.w. to achieve unaltered storage of and quick access to all these native 3D / 4D volumes.

The device setup for IVS - DVA for the entire recording system thus includes a 3D / 4D scanner and an acquisition computer, ie a powerful PC with 3D sensor, its receiver on the volume transducer or matrix transducer is attached at a defined location. For the generation of periodic-dynamic volumes, if necessary, a trigger device can also be added for timing the recording or time allocation of detected volumes. A graphic can also be seen in Fig.12, which is also valid here. If these 3D / 4D scanners have "streaming" data transmission, this would be the method of choice, but the current state of the art is rather an intermediate storage of recorded volume sequences in the scanners, which are then recorded in the scanned image Accordingly, the acquisition system must have an acquisition control and a method for the temporal allocation of the acquired volumes to the simultaneously recorded spatial coordinates of the 3D sensor.

The 3D sensor consists of a PC card which, in conjunction with its driver software, already performs a certain amount of data preprocessing. On this plug-in card is a transmitter that generates a magnetic field hemisphere, and a receiver (about in sugar cube size or in smaller variants) that registered this magnetic field, connected and operated. The data thus generated are output as a three-dimensional reference coordinate system X, Y, Z and, in turn, as a three-dimensional location coordinate system X _x F ₁ Z ₁ - that is to say in practice that the sensor generates a basic data system in which the spatial position is then registered - with others Describing words, this means that both the movement of the receiver in space and the orientation in space is detected. Engineers may call this 6 degrees of freedom, mathematicians 6 dimensions.

So far, the 3D sensor is actually a spatially 6-dimensional sensor. This is imperative for proper "tracking" of the transducer - unless you can use sensor systems that have a fixed reference coordinate system, where you would get along with 3 degrees of freedom, which will be discussed later - but first to the current State of the art with magnetic-field-based sensors:

The magnetic field has a working area diameter of typically 60 cm (hemisphere) - in this hemisphere the defined accuracy is guaranteed. The 3D sensor has a max. Error of 1 mm, which is so close to the ultrasonic resolution, so that one can actually start from a principle geometrically correct method. The ultrasound itself is no longer suitable for significantly higher resolution powers, so that fully sufficient geometric measurement functions (always incremented in mm for ultrasound) can be used, and the optimal imaging results (eg when calculating transitions from one to two) Volume to a neighboring volume) are adequate to very good according to all preparatory work or provide a good geometric basis for fine adjustments. Incidentally, such detection systems for 2D transducers have also been CE certified, namely, medical device class 2a, ie, like an ultrasound device, which indirectly substantiates this finding - and there is no reason to assume that this implied assessment of sensor accuracy would change by using a 3D transducer instead of a 2D transducer, because this does not change at the sensor absolutely.

The transmitter can simply be placed next to the patient in sufficient proximity to the receiver, which in turn can be attached (and removed) to a small base fixed to the transducer as a "clip" at a well-defined location. There are no significant magnetic fields in the room where the picture is taken, which affects the 3D sensor function, but a measurement function is available and experience has shown that this only happens in exceptional cases 3D sensor or transducer now calibrated - the geometrical properties of the sound space and its image are measured, defined and stored for all penetration depths of the transducer used, meaning that the system is fully parameterized and calibrated - and therefore basically ready for work then concerns the method of Aufna hme or their processing, which is subsequently also in a generally descriptive, mathematical form.

Both methods are described separately here. Both methods have advantages and disadvantages, as far as theoretically possible in advance. The HUMR description has already dealt with this briefly, but more with a focus on computer technology.

The PVP method described here gives an indication of where, for example, practical problems could arise with DVA: Capturing with a transducer is very different from the typical real diagnostic sounding: When capturing, the transducer is largely kept in the "average" / Angulation ", namely approximately orthogonal to the skin surface - in real diagnostic sounds, however, one performs a few sweeps, but tilts and Turns a transducer predominantly - which is then understood in virtual sounding then of course. If you "tap" when tilting or turning a neighbor volume - or is the transducer in the transition region between two or more adjacent volumes, this would then cause the idea of loading such neighbor volumes, which may be a "jump-jumping" virtual sectional image would draw.

In IVS, on the other hand, e.g. to clarify to what extent generated meta-volumes pose problems in terms of storage capacity, especially for RAM. In addition, it should be assumed that post-processing will take place after the acquisition has been completed, and one should assume that "mosaicing", ie joining together, will require a small amount of post-corrections. In addition, then probably further problems should be solved, how mutually suitable volumes at the same time with respect to their entire "projection" or positioning under the sonic model surface (virtually the virtual patient skin) can be fitted - so there is a " Matching task at least in two ways. The capture with DVA could be noticeably more complicated than with IVS, and then to capture all the Anlotungen / Anwinkelungen, so that variances between the charged volumes are kept as small as possible, so to speak, if necessary, the virtual cross-sectional image "calm down" - what possibly the load limits of However, IVS also makes considerations for a similar approach obsolete.

Many of these problems will be unanswerable or solvable without practical experimentation. Thus, there is no way around "application research" to assess the two approaches in a conclusive and well-founded way, or to decide one way or the other - or to seek and realize an optimal synthesis of both approaches Although it has been started for some time - as far as promising - but they have not yet reached a point where a first process implementation can be presented here - for a related first provisional determination, the topic is too complex or the current time still a bit too early.

For the sake of completeness, another method is mentioned here: There are now image processing processors that are able to track individual pixels in ultrasound recordings - such as in the SIESCAPE method of the company SIE. MENS. These processors are now used in some modern ultrasonic scanners. Basically, this is a very interesting approach, which could be used in principle for the generation of metavolumina - to what extent this application for 3D or for the production of metavolumina is operational, could not yet be clarified conclusively, also one would in this way This does not depend on individual manufacturers - the integration of other HUMR methods is also not available - and the recording technology for the purpose of simulation should, from today's perspective, remain independent of the manufacturers of individual ultrasonic scanners in order to be universally applicable. Basically, this feature would be realizable with very fast computers but also by software.

However - a pixel / voxel-accurate representation of meta-volumes is typically not necessary / expedient in view of the real biological variances in the training (!) - the accuracy of today's 3D sensors is sufficient for the simulation.

2.1 IVS: Integrative Volume Sampling - Creating Meta Volumes

The goal of IVS is the generation of meta-volumes. The requirements and more precise objectives were already explained in the introductory chapter "2 IVS - DVA".

A basic idea with IVS is that, in principle geometrically correct, volumes are joined together - and only around the area of a following neighbor volume according to the movement of the acquisition transducer or its sonic space - or vice versa, that the respective last volume is the corresponding preceding one It also always replaces the non-overlapping area, which one can simply decide pragmatically on the basis of the better result.

Further, in a sweep that achieves dynamic structures to be detected, trajectory pauses of the transducer may be inserted until the respective volumes (eg, at least one periodic motion are persistent) are completed, and then continue the acquisition sweep. The sweep speed is almost arbitrary, as long as it is not the size of a volume per "1 / volume rate" (volume rate = Scan speed of the transducer) on the sweep path. The sweep direction is also arbitrary - but with the practical restriction that the sweep movement must have detected each area of the desired detection area at least once in order to be able to fully generate the desired meta volume. In this case, the author can support a graphic auxiliary software (eg similar to "eraser track" in drawing programs), which visualizes how much has already been recorded or what has not yet been recorded.

For transducers with motor components, magnetic shielding of the 3D sensor on the transducer may be required to prevent magnetic interference.

In the case of 3D sensors (with the aid of cameras, in the case of optical sensors), a geometric "calibration" using anatomical lead structures, which is "repeated" for the localization of the volume in the sound model on its analog lead structures, is used to simplify the localization ( also referred to as positioning) possible. Furthermore, such a uniform recording situation - i. a uniform arrangement (position and distance) from patient to 3D sensor transmitter - realized.

The software provides the following two basic recording modes:

From a first (start) volume, the next partial volume generated by the translation (a possible small rotation should be neglected here) is added to the start volume without the overlap, this added volume will again be like the start volume as the translation progresses the eduction of the next volume increases, etc.

As before, however, the respective translational volume "overwrites" the overlap area of the previously acquired volume.

In addition, more complex addition forms are conceivable, which, however, are not further elaborated here and should be reserved for process realization. In any case, all individual volumes remain at the same time initially stored in their entirety in order to be available with all data for corrections of the "matching" (see below). After this basic detection, a "post-processing" - a so-called "matching" - takes place in which the detected volumes are matched with algorithms such as cross-correlation on the one hand with respect to their "neighborhood" - and on the other hand with regard to their "fit" for their position in the sound model, that is to say, they are primarily adapted to their correct orientation under the "skin" (or other anatomical structures) of the sound model - with this preprocessing, therefore, an optimal compromise between these two criteria must also be determined for both adaptations. in relation to the image :) "post-processing" algorithms (or "pre-processing" based on the simulation) variances of the transducer clipping during the volume acquisition (see also part 6 and 8) can be post-corrected.

Concluding Practical Note: If you know the Freehand method and understand the above capture process and then compare the two, it becomes clear that IVS capture is likely to be drastically simplified - and that, even with larger volumes, will still have artifact freedom and Improve resolution.

As detailed in the following chapters, to achieve sufficient "perfection" of the simulation, it will be necessary to record equal volumes multiple times (eg from different body regions or "leads") - nevertheless, this "additional effort" may be due to the inherent low Artifactivity and the ease of use of the acquisition software (see "Eraser" - Tracking the recording) in IVS-DVA make this extra effort bearable.

Below is the mathematical description of IVS:

Preliminary remarks:

• The most important characteristic of this method is its ability to adjoin several neighboring, scanned sound volumes not only in an approximate (and computer-assisted) way, but in principle geometrically correct. "Basically" means that in the realization of a process system-related inaccuracies are taken into account, and in the biological medical reality when scanning artifacts such as small shifts (caused by respiratory movements, etc.) of the body structures can arise from one volume to the next, may require the additional "post-correcting" optimization procedures. The- nor does this process have precisely this specific ability, as a first step, to provide a geometrically correct, defined basis for such an association of sound volumes. (This definitely has a practical significance in the degree of "correctness" for the reproduction of physiological / pathological structures, if not only "overview / orientation", but concrete diagnostic differentiations are required, and in particular also in a subsequent application of measuring functions. In contrast to an ultrasound scanner used only for immediate diagnosis - here acting as a central recording component - the W / ectergrabe system "simulator" which is also under discussion does not have the possibility of obscuring it by rescanning proportional volumes ( of a "meta-volume") on the patient, ie the captured and reproduced in the simulator "MetaVolumen" then virtually replaces the real patient and should therefore provide all the diagnostic information necessary image information in the highest possible (geometric) quality immediately increased picture accuracy In addition, the use of ultrasound scanners for purely diagnostic (ie not as a recording system for the simulation) has given a medically relevant benefit to "meta-volume", in principle simply because the (geometrically) correct image / volume Data can be made available even after the examination or presence of the patient by saving.

• At this point it is important to distinguish between the following general or "principled", albeit mathematical, description of this method on the one hand, and one or more concrete implementations of this method on the other hand which is not (yet) to be done at this point, so here is a basic mathematical description of this method.

output data

A (A.) 3D / 4D ultrasound scanner described in the introduction with the functional principle, a (B.) image and transducer localization detection system connected therewith (currently a PC with corresponding soft - would be) with a likewise described there (C.) 3D sensor (6-dimensional) used. The immediately as 3dimensional space the sound data generating device of such a system is the belonging to the 3D / 4D ultrasound scanner transducer. Its working area, where in the form of so-called volume the sound data are generated, is also referred to below as sound (values) space. Scanner, about 3 to 5 volumes / sec. are generally referred to as 3D systems, faster scanners with about 20 volumes / sec. For example, for the real-time representation of the moving heart - as 4D systems (inclusion of the time axis), and possibly even higher dimensions could be cited by means of color: Here, therefore, quasi "at least three-dimensional scanners" are relevant. A., B., C.) can in any case be separate units of equipment or an integrated system and are referred to in their entirety as a receiving system referred to for the IVS method.

With the help of the recording system will be at the times

0 </, ≤ / ₂ ≤ ... </ ", n ≥ \

Parts of the human body (directly through the transducer, which is acoustically coupled to the patient) sounded. At each point in time t _t , V _{1 designates} the corresponding "sounded" volume.

The sounded volume is not considered here in the form of raw data, but has already been transformed into the output coordinate system. This could be, for example, a cuboid, a truncated cone or a pyramid. A typical example of a so-called matrix transducer is the 3D / 4D scanner of interest here, for example a (regular, straight) truncated pyramid. In these explanations, a (regular, even) pyramid was chosen as an example for all transducers that generate a (sound) volume. All these geometric objects can be parameterized (which is also necessary for their practical implementation in corresponding algorithms and will then be described): The (regular, even) pyramid, for example, is clearly indicated by its peak (as a three-dimensional point) the solder is fixed to the base (as a three-dimensional vector) and the base surface length. In general, parametrizations can be described by specifying a parameter set Θ and a parameter θ e Θ. In the case of such a pyramid that would be eg Θ: = R ³ x R ³ x (0, ∞)

In this case, the sound volume is parameterized, i.

K: = V (θ (t ")), 0 (O e Θ

where Θ is a suitable set of parameters. The sounded data are through the

Functions f ": V _n → Ω, l ≤ n ≤ N

specified; Ω designates the sound volume space.

Sound volume refers to the set of all possible gray, color or other "intensity values" or "sound values" that the ultrasonic scanner can "return" for the individual coordinate points, f, (x) is then the value the time

/, sounded at position x.

Method for static meta-volumes

The statically oriented method is intended for structures of the human body that undergo virtually no temporal or periodic changes (as is the case in the heart or respiration). It can be described mathematically simpler than the dynamically oriented method and thus has preparatory character.

In the statically oriented method, the sound volumes are joined together one after the other. The overall picture consists of the sounded data for the individual volumes. For the functionality of the method, the sound order in the basic principle is irrelevant; This could also be reversed or selected, etc. In addition, it is not necessary to use all sound times; For example, only times at the beginning of the scan could be considered. Generally we specify the selected times and their order by the (injective) function α: {i, ..., A:} → {i, ..., N}

where l ≤ K ≤ N and AT indicates the number of times selected. The first sound time is then given by / _{α (1)} , the second by / _{α (2)} and so on.

Injective means that α (i) ≠ a (J) for all i ≠ j, i. that each volume is considered at most once.

If we restrict the overall picture to a user-selected area V <z R ³ , we obtain the following sound range:

VV _{a (k)}

Since individual volumes can overlap, there are several sound values for each coordinate point based on their intersection. The statically oriented method describes, for each coordinate point from the sound area, which sound point is to be used to determine the associated sound value:

j: V → {1, ..., K}, j (x) = minjt: 1 <k <K, xe V _k }

With this picture, finally, the composite sound image through

in a clear way.

In the simple case K = N and a (k) = k, the sound / intensity value of the sound time point for which the position x was sounded for the first time is used for x.

Method for hybrid static + dynamic meta-volumes

The procedure for hybrid static + dynamic meta-volumes considers biologically relatively frequent movements (respiration, circulatory pulsations, etc.) of sound structures to be detected. These are usually (1) repetitive / periodic and (2) "regionally limited", that is, possibly only parts of an entire region desired for detection.The principle is based on the above-described statically oriented method.

The desired scan movement of the physician during the volume acquisition, the so-called "sweep", can be described as follows: In general, it is fluent In between, the transducer of the scanner is practically not moved to avoid a moving structure. a part of it) from the same position to scan over at least a complete period.The sound data of the volumes that belong to the individual stationary transducer positions of the scanner are to be summed up again, this time with synchronization of temporal changes of the moving sound structure (eg One could probably reduce the scanning process for moving sound structures only to a moving structure - at least if the sound space of the transducer can completely map this moving structure - but this is logically a subset of what is described here and therefore chosen method is because even the Abd Eckbarkeit the largest possible volumes is desirable.

Let K be the number of volumes that you want to "sum up", ie the number of rest positions that should be used for the dynamically oriented procedure.

T denote the period length and

s _m e [0, T), λ ≤ m ≤ M

the times at which sound data should be synchronized. As a rule, one will choose s _m equidistant, that is s _m = (m - ϊ) - T / M. To describe which ones

Scan times for the procedure (eg which exact period during the k-th row position) are used, we again use the above function a, this time extended by a time-dependent factor (where we do not use the time s _m but instead to express by means of the index m):

a: {l, ..., K} χ {\, ..., M} → {h ..., N} At the time of sound t _{a (km)} , therefore, sound data from the k-th rest position are present, which are to be evaluated for the synchronization time s _m . The above-demanded injectivity can be replaced by the assumption that the quantities

A _k = {a (k, m) Λ ≦ m ≦ M}, l ≦ k ≦ K

are disjoint, so do not overlap in pairs. The choice of the function a naturally influences the order of accumulation; it must be defined either manually or automatically. In this general formulation, M need not necessarily coincide with the actual number of sound images that the scanner can deliver based on its frequency within a period. However, in practice this will probably be the case and one will also choose a scan time for the k-th rest position - for example: /, which marks the beginning of a complete period, and then a (k, \) = n , a (k, 2) = n + la (k, M) = n + M - l.

The scanned volumes during a single resting position are nearly identical (except for any fluctuations, such as the patient's hand shaking or other discontinuities in the detection sweeps), i. it applies

V ^γ a (k, \) ~ ~ V ^y a (k, 2) ~ ~ ^{■ ■ ■} ~ ~ V a (kM) '1 <Jr <K

To describe the composite sound image well-defined, we set

v _k = (y _{a (k, m)} , \ ≤ k ≤ κ

and ignoring any fluctuations. For each xe V _k , therefore, there is exactly one sound value f _{a (km)} (x) at each instant s _m . If we restrict the overall picture to a user-selected area V cz R ³ , we obtain the following sound range:

With the function j: V → {l, ..., K}, J (x)

: l ≦ k ≦ K, xe V _k }

(which is identical to that from the statically oriented method), the dynamically oriented method then passes through

/: V x {1, ... M) → Ω, f (x, m) = f _{a (j (x) m)} (x)

in a clear way.

Final remark:

It should be pointed out once more that the general principle and quasi the "innovative essence" of the IVS-method should be described here in a general way.Additional practical optimization methods (fine tuning for a matching / mosaicing and the like) should be described in this context or They are certainly sensible and necessary, but may change relatively quickly and should be explained in terms of their own representations of this principle, and again in more general terms: the performance of the ITS procedure lies not only in a simple "mosaicing" (ie, joining without gaps or overlaps) static ultrasonic volumes, but, more or less two steps further, in that, during a single seamless, possibly (multiple) pausing, but actually continuous "sweeps" captures "mixed static and dynamic" volumes as "mosaicing" captured / generi The result here are so-called (hybrid) "meta-volumes", in which static and dynamic (meta) volume components are virtually arbitrarily mixed or "hybrid" exist - and this in a basically geometrically correct form. The immediate visualizability of a whole or even "half" patient is not really effective in teaching: Ultrasound is divided into many disciplines: For example, those who are learning abdominal ultrasound do not want and need the immediate visual availability of the breast (female breast), who echocardiography ( Heart) does not need the immediate visual availability of the genitourinary tract, etc. - and this is not only true for a highly specialized pathology training for experts, but even for a first orienting sono-anatomical (basic) education The immediate visual availability of the direct "real" environment of an ROI ("Region Of Interest") to the presentation and training possibility of the use of so-called anatomical lead structures ("Landmarks"), on the one hand the conceivability of the given morphology of an ROI together with the relevant environment makes it possible On the other hand, there are practical, often necessary aids for finding the sound object together with the correct transducer setting (slotting) or cutting plane. A physiological (ie normal-healthy or other) "standard patient" should be considered in this simulator only for basic sonographic anatomy, otherwise "environments" should be patient-specifically visualized and correspondingly realistic. Making this possible is one of HUMR's features, and these are the objectives of HUMR. More distant environments or volumes can also be completely realized by selection (ie in terms of visualization speed via the GUI menu or targeted loading of the desired anatomical subregions instead of permanent provision in the RAM of the system). HUMR, for example, does justice to multimorbid patients and their simulations - even several selected disjunctive regions can be realized in this way. Perfections of the simulation are to be focused at HUMR on aspects that are actually relevant to the training (ie "simulation sufficiently realistic, as ultrasound in the specific cases really looks like and how well trained or training is practiced"), but there with some Effort, as far as it is necessary.

The performance of the HUMR method (s), in turn, is to couple and integrate the IVS method so seamlessly with associated ancillary procedures that the detection of such (hybrid) "meta-volumes" is in fact essential for all diagnostic ultrasounds that are essential today -Modalities in practice and can then be visualized by "downstream" simulation technology - which is a sufficiently methodologically complete and probably as sufficiently realistic-looking or one according to qualified and qualified training sufficient ultrasound simulation actually only (once) possible (at least according to the - presumable - judgment of the vast majority of experts, but which is the best possible attainable reference or real scale.

Corresponding highly qualified reports of eg the previous president - Prof. Gebet, recognized top expert for ultrasound as well as its simulation - DEG U M. ^¬ German Society for Ultrasound in Medicine - Germany is at least one of the leading nations in practical ultrasound including training -, This, as well as a concept assessment, could well be provided, if this is relevant here.

It is in the nature of the matter that "HUMR-procedure" is also a plural (even in a final patent application, it is not clear): A comprehensive description is only as here, verbally explanatory possible, and even a general The mathematical description (eg for IVS) can only be represented in some parts or not over all and really meaningful.The nature of the thing is just that here completely different ultrasonic problems are solved, which have only one common denominator, and this is the comprehensibility of sufficiently accurate imaging and simulation-suitable ultrasound (meta) volumes - with which 3D / 4D or other auxiliary method (mechanical, electronic, PC-based), be it "providing and supporting scanning and simulating scanning processes" or The reason for this was already mentioned in the introduction: The non-existence of an ultrasound scan rs, which can capture the entire patient, and this in all scan modalities, even with a single scan - ideally for static body parts in a split second and for periodic organ movements almost or practically simultaneously recording a movement period - requires one (also multiple) delayed replication in all relevant examination scenarios and scan modalities for the visualization of all ultrasound examinations in the simulation. The simulation of ultrasound is in principle a task - but it can only answer for the time being with many answers: This is also a comprehensive description possibility of HUMR.

For the rest, however, the detection of such meta-volumes may also be without one

"Downstream" simulation (but possibly with slightly modified visualization technology, which is initially only available with the help of simulator components such as patient Model and transducer imitation was realized, but, for example, by a mouse control, could be replaced), in a similar form already be relevant for immediate patient diagnosis or possible therapy. This may have limitations with pronounced interactive sound techniques such as gastro-enterology or emergency medicine - in gynecology on the other hand, this is quite conceivable - and this option gives the ultrasound a bit of possibilities, as they had previously only radiological procedures such as CT or MR, where a patient can be scanned in whole or in larger regions - and can only be diagnosed later - or earlier scans can be compared with later in such a way not previously given way.

To structurally build this script:

Such a mathematical description has according to previous knowledge only for the above IVS method some sense - possibly even an analog Oschreibung for the following DVA procedure is created. In the other methods of this document, such a generalization by means of mathematics is not so representable or appropriate. Therefore, their presentation, where appropriate, for the time being only with the means verbal (technical and functional) descriptions, technical drawings and illustrations or explanation of software such as algorithm description and flow charts.

2.1 DVA: Direct Volume Access - Pretending Meta Volumes

An alternative approach to IVS in the first approach is DVA (Direct Volume Access).

This method is in principle a variant of the PVP method, which is explained in more detail in the following Chapter 3 - the difference is in principle only that PVP is based on meta-volumes, whereas DVA bases this approach on the underlying individual volumes applies. Further details are described there. The description of DVA is therefore quite short to avoid redundancy.

DVA is thus basically very simply describable by the fact that individual - typically adjacent - recorded volumes for simulation (after localization in the sound model) correspond to the instantaneous position of the transducer. If this is possible at a sufficiently high speed, the effect of a meta-volume or its size is visually achieved in principle - some of which may or may not be necessary Auxiliary techniques are further explained in the following chapters 5 "SSP" and in particular 6 "VTP".

A prerequisite for the DVA method is a sufficiently high transfer speed of the storage media (disks, etc. - RAMs are fast enough anyway), on which the individual volumes of the region to be visualized are stored. This can be assumed in principle meanwhile.

As mentioned in the introduction, practical research work must show whether IVS or DVA is the more appropriate method - from today's perspective, a combination of components from both approaches seems most likely. This also depends on the available computer hardware, operating systems and other ancillary technologies, which will be discussed in the final chapter 10.

3 PVP - Perspective-Related (Meta) Volume Processing

(Meta) volumes that are from a particular anatomical attachment point of the capture transducer, e.g. "From above" (as seen here as an example from the abdominal wall) can typically not later be correctly visualized / simulated by the flank (i.e., ventral side) - but this is frequently practiced during sounding.

On the side of the (meta) volumes captured, eg "from above", there may be structures that are located in front of the volume in a different perspective (eg flank or ventral side) but not captured "from above" - or even then when they have been captured, by which the other direction or perspective is at least visually altered - and thus generate a different, distorted image, if the perspective or visualization in the simulation does not correspond to the perspective / visualization during the recording. Likewise, some physical phenomena of ultrasound, such as sonic shadows in detected volumes that are mapped along the propagation direction of the sound waves, are no longer correctly reproduced in the subsequent simulation in the case of other plots than in the acquisition - eg. For example, sonic shadows that are imaged perpendicular to the transducer sound surface during recording, are no longer perpendicular to the "sound surface" of the transducer imitation of a simulator during playback / simulation that does not correspond to the recording sound previous provenance.

An at least approximate solution to this problem, which can reduce such errors to a no longer disturbing Mali, is achieved with the PVP method: PVP is or requires the multiple recording of a (meta) volume from several directions - at least as far as they are diagnostically relevant or appropriate for a sufficiently realistic reproduction from different approaches and can then be visualized accordingly in the subsequent simulation. For this purpose, an initial sounding is selected during the recording, to which the following individual recordings apply. The individual (meta) volumes recorded in this way are then positioned analogously in the sound model. The image processing component (s) of the simulator now store all these individual volumes on the memory side in such a way that they can be visualized immediately when the transducer imitates a corresponding position or sound.

For this and related problems, multiple (meta) volumes need to be captured from "relevant" (ie, substantial difference generating) perspectives (= transducer positions) and made to be quickly accessible in real time to the respective position of the transducer imitator on the simulator to adjust.

The necessary prerequisites for the interaction with the sound model and the transducer imitation to provide the system with the necessary control information are described in more detail in the methods explained below under Chapter 5 "SSP" and Chapter 6 "VTP". These methods have all the data available through their comprehensive depiction of the geometry and dynamics of the sound model surface and transducer fidelity in order to determine the respective positions or approaches. In the case of the transducer imitation, the perspective (meta) volume can be loaded or visualized.

If this proves necessary and useful, PVP may even mean that volumes are acquired from the same "perspective," but with different (per "sweep" but constant) angulations of the uptake transducer - but this only makes sense if different angulations from otherwise anatomically identical Aufsetz-areas of the receiving transducer or the same or similar sweep lines actually diagnostically or visually significantly significantly different or different images to generate.

Further perfecting can be achieved by using the same exploration-specific acquisition technique described in Chap. 8 "ADP" for Doppler, which is also used here for PVP and also used for gray-scale images for the later exploration-related retrieval or the visualization / simulation of analogue exploration variances in the simulator's transducer imitat. When this makes sense can hardly be formulated - you will have to decide on a case-by-case basis when to rely on it.As a standard recording technique, this will be especially in image capture with 3D / 4D transducers, which are likely to lead to orthogonal Anlotungen, and also probably not because of the not incommensurable borrowing effort.

The DVA method described above applies this technique to individual, typically directly adjacent (single) volumes rather than meta-volumes.

4 DVT - Dynamic Volume Triggering

For the acquisition and visualization / simulation of moving organs / structures, triggering systems can be used, as far as cyclic / periodic / repetitive motion patterns are concerned.

Triggers form, so to speak, a time-clocking, in order to enable not only a spatial assignment of images, but also a temporal assignment - for example, to sum up several images / volumes with a sufficiently good time resolution. The purpose of such a timing is to generate a time grid, by which output images (in addition to the spatial assignment) provided with a temporal assignment to sum them up until the result images have a sufficient number of pixels / voxels for all periods of a motion period (eg, a heartbeat). After acquisition, the result can be visualized as a "loop" of a period of motion, but real-time scanners can often render their trigger systems obsolete due to their high image / volume frequency, since such a "summation" Several triggers can also be used in combination.

For movements generated by the heart (immediate heart muscle, valves, etc., or vascular pulsations indirectly generated by the varying pressure or flow conditions of the blood), e.g. ECG triggers are used - the following explanations on echocardiography are in principle technically applicable with regard to triggering to the vessels connected to the heart - further aspects of angiology / phlebology are dealt with in chapter (8) ADP.

A. For echocardiography:

In the representation of any 2-dimensional sections of the adult, pediatric or fetal heart, the 3D / 4D recording / presentation provides an optimal basis, since each 2-dimensional section can be calculated / reconstructed and represented - in principle, also cuts that otherwise can not be visualized from abdominal (or theoretically endovaginal) probes with B-mode transducers. In the simulation, this could also be consciously limited to simulating the B-mode if necessary - as a kind of "negative PVP", where nothing is visualized outside the physiological sound window Technique, first in the TEE (TransEsophageal Echocardiography), and then also in the TTE (TransThoracal Echocardiography) in the adult and partly also in the pediatric cardiology comprehensive use has been made for the simulation is anyway a (dynamic) volume methodological requirement Fetal echocardiography has a special status in prenatal medicine: Firstly, "heart defects" rank first in terms of incidence (frequency) among the fetal malformations, and secondly, there is hardly a malformation category where there are so many therapeutic options gives. Although many gynecologists may respond to heart defects with referral to specialized colleagues - the prerequisite for this is yet to recognize such a malformation, so this is due to the very high training requirements in the light of his medical Relevance should change little. (The task of the specialists can not be such a basic stratification of all fetuses of the nation, but a "fine diagnostics" following referral, possibly initiation of a delivery at an opportune time, coordinated preparation of rapid therapeutic postpartum measures by the pediatric cardiologist and the same.)

Technically, however, fetal echocardiography has a serious problem: For cardiac ultrasound images / volumes, a trigger procedure is needed. In adults, this can be easily resolved with an ECG. Since an ECG is not an innovation, it is only mentioned as a matter of course in the patent application. In a fetus, however, only a "nonsensical" ECG is derived, and this reflects the heart of the mother - is thus therefore unusable (a fetal heart works practically independent of the mother and incidentally has a typical about twice as high heart rate as adults - but It is also quite unsteady: a separate trigger procedure is indispensable.) Now, fetal Doppler ultrasound as the only practicably available alternative deriva- tive seems to offer a first obvious solution - also the Doppler flow (-Shift) tips are similar to one R-wave is technically easy to define on the ECG - but ultrasound imaging and fetal doppler ultrasound massively interfere with each other's signals: when both are simultaneously applied to the same sounding object (fetal heart), both are no longer usable is the background against which you have to see the acquisition of useful dynamic volumes - or just barely got to see - except for "Sunday recording n "of fetal hearts with a very constant heart rate. Primary interest for the simulation, however, is precisely pathological hearts, where one can not assume that "ideal" or long-term stable heart rates are involved.

Do you even need a fetal echocardiographic trigger?

The next obvious idea might be to use real-time 4D scanners with matrix transducers that do not perform "volume reconstruction" because the transducers generate native volumes and are known to be so fast ("real-time 4D"). in that they do not need an ECG trigger for image generation. (In order to be able to exactly determine the assignment of the images to the cardiac phases, however, an ECG without triggers is nevertheless implemented). They are expensive, but if money does not matter, it seems plausible at first. However, It is well known that these systems tend to rank in the geometric resolution at the lower end of the ultrasound scanners. Here it should be further clarified: For each penetration depth a certain number of "voxels per unit volume (eg cubic centimeter)" for the geometric resolution is technically specified or defined - in other words: A large heart gets many voxels, a small correspondingly few voxels for the geometric resolution: These matrix transducers were designed for (possibly pediatric and) adult echocardiography - and the adult heart can have a "tens of times" larger "cubic centimeter or milliliter volume" (or muscle mass plus cavum / blood volume - that is the volume to be visualized) of a small fetal heart reach - correspondingly few voxels are available for the resolution.A per se, a fetal heart is already "roughly" dissolved - and the lower the physically-technically predetermined resolution of a Procedure, the more visible on top of that. In addition, the child's or adult's heart is also much closer to the transducer (starting directly behind the thoracic wall or visualizable from the sternum) than the fetal heart, in which first a sound path from the abdominal wall or placenta of the mother through the amniotic fluid If the penetration depth reaches, we may already be in the "far field" or in the vicinity of it, where the geometric resolution then decreases again. "How fast is" fast "with matrix transducers? ? Currently about 20 voices / sec - and thus back to the fetal heart rate: it is typically about twice as high as that of an adult heart. . Assuming an adult heart rate of 60 bpm of a heart action in 1 sec is dissolved with 20 volumes of time by assuming that a typical fetal heart rate of 120 bpm, this is dissolved in ^Λ A sec with 10 volumes -. So in Average for the 4 phases: Early and late systole as well as diastole each 2.5 volumes as movement or temporal resolution - not too much, so relative is thus "fast".

Theoretical considerations suggest that the results of a matrix transducer may be useful, but will not herald a "new age." The results of any future special matrix transducers that will optimize for these particular fetal-echo constraints Here are back to the money: Only for fetal echocardiography to buy an expensive real-time 4D scanner, with which one currently could do little else in gynecology, is also disproportionate for specialists: If the result a Would promise a "new dimension" - perhaps for a few, but that is not to be expected for the time being.

Back to the initial question - with the answer. So, according to the current state of the art, a fetal echocardiographic trigger really makes sense:

The fetal heart is one of the cases where the presentation of a larger anatomical environment for the training can be done relatively well - the transitions to the large vessels close to the heart, such as the aorta, etc., are also visualized, and at least from birth on Airborne lung anyway no longer much relevant to see in the ultrasound. The Freehand procedure therefore does not have to complete long sweeps for the fetal heart and works with 25 2D-images / sec. or analogously just as many reconstructed volumes rather faster than matrix scanner. (Possibly also the so-called "Halbbild-Verfahren" can be used for training purposes: Since the data transfer takes place by video, here according to the TV standard two consecutive pictures with the lines "1, 3, 5, ... to 767" and "2,4,6 ... to 768" before, just the so-called fields - these can be converted by interpolation in each case to "frames" and thus achieve a doubling of the frame rate to 50 / sec - here would have to decide accepted experts, whether image quality is such for education that it can reasonably be said "that's how it might look like in reality".)

The first prototype of the FET (Fetal Echocardiography Trigger) was tested with this method and showed the expected results, as long as the simulation performed cuts in approximately the same plane as when recording with the 2D scanner - with more orthogonal sections left the Picture quality strongly after - which did not surprise however so also. (The field method has not been used / tested.)

For the simulation, a case should therefore be detected in two mutually approximately orthogonal sweep directions and made available at least analogously in the simulation with a given cutting plane specification. However, this limits the perfection of the simulation - however, this is sufficiently solvable that the simulator loads these two 4D volumes and visualizes the "matching" 4D volume by determining the position of the transducer (see also "PRS - Perspective Related scanning "). Incidentally, there is also an interesting alternative: When using the FET in a 3D scanner, the effect described above (lower quality at other cutting planes than the one used for recording) is not or only to a much lesser extent expect. The FET can be used virtually unchanged for a volume addition over the time axis, for which you can then use conventional 3D scanners with volume transducers, which are designed for abdominal clutches of a fetus, and otherwise can be used by the gynecologist. Thus, the FET can be used as an alternative to matrix scanners with two scanning methods, which is quite reasonable to evaluate.

As a routine procedure, FET is currently not yet conceived (but could be further optimized if necessary), but as a solution to capture at least fetal echocardiographic recordings in selected cases for simulation purposes. So much for the background of the FET.

To describe the FET method itself:

Since the Doppler sound with the imaging sound - as well as the imaging sound with the Doppler sound - would produce massive interferences (both processes could then no longer deliver usable results - a statement that also confirmed preliminary tests) used a time-shifted method.

Thus, this method optionally utilizes a separate ultrasonic Doppler device associated with the triggering system or the Doppler function of the transducer of an ultrasound scanner to determine the fetal heart rate over a representative period immediately prior to image acquisition. In detail presented here is the

I. Version with separate Doppler device, 2D scanner and the Freehand method:

The trigger device FET (Fetal Echocardiography Trigger) developed for this purpose evaluates the audio signal of the Doppler in accordance with the desired cardiac fetal rhythm as peak flow at the beginning of the ejection of each (periodic) heart action. This signal is easily recognizable / discriminated because it has a much higher amplitude than the remaining Doppler signals. Figure 10 illustrates this immediately: The rhythm determined in this way corresponds to the heart rate, which is transmitted as a TTL output pulse to the image acquisition system for 4D reconstruction. As a check, the display shows the current output frequency.

The output frequency of the trigger device could be comparatively validated by manual remeasurement of audio tapes with the aid of signal evaluation software.

The trigger device in the strictest sense consists of an analog component, which uses a bandpass filter that filters out interfering frequencies, and a microcontroller that uses an integrated A / D converter to measure the Doppler audio signal according to a practically proven algorithm. The device synchronizes its output signal when the Doppler signal is applied. If there is no input signal (Doppler switched off for image acquisition), the last measured frequency is output as a "virtual pulse".

A tracking / updating, intermittent adaptation of the "virtual" pulse in the case of a "divergence" from the real pulse - i. with a brief pause of image acquisition - is provided, but was not necessary in the cases recorded so far. The virtual pulse determined in this way takes on the function of the ECG trigger as it does for generating a 4D image sequence in "conventional" cardiology. The immediately determined trigger time can be provided with a temporal shift according to pragmatic aspects of optimal image acquisition.

This makes it possible for the software to perform the necessary or even the most suitable temporal arrangement of the 2D images of the fetal heart before conversion into a moving 3D volume or a 4D volume. The result is now a sequence of correctly consecutive volumes at intervals of 40 (original data frames) or 20 (so-called interpolated fields) msec. or 25 or 50 images / sec. for the duration of a fetal heart action, which then generate a dynamic 3D volume or a 4D volume. So far, only the full-screen method has been tested.

The trigger pulse which is initial for the image sequence can not only be "preset" by parameterization of the trigger system, but also according to a "life image estimation" of the observed heart action immediately prior to the acquisition, even in the case of the The recording system (MedCom ScanNT) can be set at a theoretically arbitrary position.

In practice, it is advisable to first select the trigger point in such a way that it is set in a longer, low-motion phase (eg in the end diastole, but sufficiently far in advance of the contraction) - this way, one can "jolt" when "playing" Avoid when images that are just captured on the border between two consecutive cardiac cycles are placed "once as last" image of the one cycle, and "times first" of the next cycle: in both cases, they are considered late-diastolic images in the designated area, this is immaterial for the visual contraction-dilation process - however, if one or the other early-systolic image is classified in the late-diastolic area, the result is medically no longer correct and acts when "play "even very visually very annoying.The screenshot below shows the acquisition of a heart - bottom left to right In the central window "US Acquisition" you can see the symbolized R wave, the set trigger point, the already sampled frames, the heart rate and their allowed variance (Fig. 11 by courtesy of MedCom GmbH).

Figure 12 shows synoptically and synchronously again the sequence of a heart-triggered with FET (Doppler: blue) or ECG (ECG: yellow) heart-triggered (trigger signal: shown in red) recording; diagonally from left / top to right / bottom: Geometrically not yet Cartesian transformed M-mode / 2D images (frames) ECG (for ECG triggering - in the case of FET triggering here only as an orientation for the assignment to the cardiac phases to understand)

Derived trigger signal Doppler-based hemodynamic signal

Timeline (running diagonally from bottom left to top right) Timed and geometric (see up arrow on the first "image cubes" for the spatial assignment of the frames) correctly arranged 2D images:

Finally, a Figure 13 with a schematic diagram of the entire device structure with separate Doppler + FET (left), 3D sensor on the transducer, 2D scanner (center) and recording system (right) during the recording process in the patient: (Also the spatial assignment by 3D-sensor is shown in this last graph concerning DVT-FET for the complete device-structure possibly more vividly and in addition to the above graphic).

The Freehand method with these representations was also chosen here because FET is applicable to it and at the same time visually a good idea can be conveyed, as 3D and triggered 4D "step by step" run - iw also for 3D / 4D scanner where this - except for 3D sensor and FET - scanner internally (as an integrated system) in principle very similar realized.

II. Version with 3D Scanner (Volume Transducer):

A 3D scanner is also another alternative for FET: The device can also be used to properly position / solder the volume transducer of a 3D scanner to capture the fetal heart and then, for a sufficiently long period of time, measure the volumes with "Time Stamp Instead of sweeping the freehand method, the volume transducer of a 3D scanner must be set in the penetration depth in such a way that the fetal heart is optimally and completely visualized and then recorded in this plot during the acquisition. stay positioned.

When capturing the volumes, it may be necessary to use an n-fold of the period duration of a heart action for summation, for example 5 volumes / sec, because of the too slow time resolution of volume transducers. On the one hand, a fetal heart will not dissolve in an acceptable manner, and the volumes will also be statistically better distributed, eg at 3 or 4 periods, or may be regularly distributed evenly and then regularly inserted into a period length according to the order of assignment between two trigger times , By contrast, the recording system and the simulator can process a higher frame rate (eg 25 volumes / sec.). So then a corresponding number of volumes could be summed and used and then displayed in physiologically correct "playback speed" with good time resolution. III. Procedure with Real-Time 3D / 4D Scanner (4D Matrix Transducer):

These procedures may be unnecessary for fetal echocardiography using Real-Time 3 / 4D scanners with matrix transducers: their high image / volume frequency makes these scanners obsolete "summing up" image data.

Initial attempts seem promising, but have not yet been published.

B. For breathing:

For triggering in breathing, many different methods, e.g. Impedance method can be used. Another option is a pressure-sensitive belt that can register breathing. Likewise, for this purpose, a 3D sensor can be used, which is positioned in addition to the 3D sensor of the recording transducer at a respiration-representative position on the patient. The choice of the trigger used can thus be chosen relatively arbitrarily as an appropriate solution, which corresponds to the respective question, taking into account the required or desired accuracy and the effort.

C. Irregular or "chaotic" movements:

For irregular movements - like the whole-body movements of a fetus or some movements in orthopedics tec. - Real-Time 4D scanners will produce really satisfying results.

5 SSP - Scan-Model Surface Processing

So far, ultrasound simulation has simply used volumes that have been recorded with ultrasonic scanners and 3D sensors, and then visualized by means of a 3D sensor in a pseudo-transducer "virtual", ie calculated from its spatial position cuts. The 3D sensor and the simulator system thus took into account the shape and spatial position of the recorded "data cloud" within the sound model, more or less like a three-dimensional projection medium for the recorded volume - as well as the current position of the 3D sensor. The shape, or more precisely: the (surface) location coordinates of the sound model were irrelevant or were not recorded and calculated for the software - initially logical and sufficient appearing, since the detected volume can still be localized in the anatomically correct position in the sound model or every desired cut could be visualized. This led to the effect that sectional images were generated at some positions of the Peudo transducer even if it was not set up on the sound model - a somewhat lacking realism, which in itself is not very (negative) consequential, because it is rarely noticed and even then, because of its obviousness, does not really lead to the misrepresentation of false bi-interpretations.

In other words, ultrasound simulators have only "known" the volume and position of their pseudo-transducer, but not the sound model, which is only a passive and quasi "qualitative" but "quantitatively unknown" (ie, location coordinate) surface to simulate the system component "Patient (in)". For a variety of significant reasons, here now (3-dimensional) scanning, e.g. with a 3D laser scanner, all types of sound models used (such as for cardiology, internal / emergency medicine, gynecology / prenatal medicine, etc.). (Prerequisite for the meaningfulness of this method is a sufficient uniformity in the realization of all produced specimens of each sound model type (one could also say: sound model class), if one wants to scan only the sound model type and not every single sound model, which would be unrealistic.) There are currently three technical implementation paths for such a "sampling":

(1) Using the "own" 3D sensor with a sufficiently uniform scan method to obtain a sufficiently dense "grid", the surface of the sound model approximately "orginalgetreu" (sufficiently small polygons, orthogonal angling, etc.) and everywhere "uncompressed "(in the first approach as in the untouched state, but possibly also plus / minus a small, desired offset) reproduces;

(2) using a special optical (e.g., laser) scanner that could in principle make the surface data "finer";

(3) Obtaining this surface data directly from the manufacturing process of the respective sound models. The goal is in any case a sufficient "dense" network of surface data (eg polygon corners), which ensures a sufficiently "fine" image of the round, irregularly shaped sound models. The latter two methods are expected to be generally more accurate in practical implementation, but this might require an additional computational step, which makes a suitable conversion of the acquired surface coordinates into the proprietary coordinate system of the own SD sensor - ultimately a pragmatic or in an attempt decision to make, which works best in practice.

In any case, such a scanning method makes it possible to store the surface data of the sound models as a sufficiently finely resolved multiplicity of individual, coherent polygon planes in a table, to retrieve them quickly and to use them for further processing for image data visualization and simulation.

There are few problems to be expected in the implementation. In practice, the exact uniformity of all sound models of each sound model class is likely to be one of the biggest challenges for the current relatively small numbers of series.

(Theoretically, as a fourth way, a pressure sensor / switch on the pseudo transducers would be conceivable, but in many ways would be a drastically less powerful solution - as can be derived from the following chapters.

Under certain circumstances, however, a combination of both methods - scanning of sound model surfaces plus pressure sensor may be useful for some problems.)

In particular haptic sound models (see AHS) can also be provided with further anatomical features such as a skeleton, which follows the respiratory movements. In such sound models, which can perform active or reactive proper motions, it may be necessary to perform a plurality of such surface scans, which in this case reflect desired states - and thus include the loading of the variable surfaces into the visualization or simulation. The given by these Einscan-method, only with "calculated touch" or "calculated contact with impressions" of the pseudo-transducer in the sound model now on the monitor of the simulator actually an "active" sound image (versus "blank image representation", as If the transducer is not set up, this is actually only the "most obvious" advantage of this method. "In fact, the methods of scanning the sound models in an elegant, ie very simple, uniform and computationally economical manner enable as few or as a retrieval table already existing "measurement values generate as many useful, derivable sequence parameters - the realization of many more important features, which are presented below up to and including chapter 6.

A fundamentally important step in the visualization of image data (volumes) is the "positioning" in the sound model at the "right", i. the real patients, for whom the respective case reports (cases / volumes) were recorded, corresponding localization. This can be realized by various methods - translation of the volumes via corresponding individual functions such as shifts of the individual spatial axes, rotation of the volume, storage of one or more cutting planes or defined points in recorded volumes, which are then reproduced as a reference on or in the sound model, etc. This will typically be based on a 3D sensor with 6 degrees of freedom. When using conceivable other - e.g. optical - sensors that involve a predefined position - and thus a separate reference coordinate system when recording (eg on a special recording table) or visualization / simulation (eg via optical components such as lasers, cameras and the like), may possibly affect 3 of the otherwise 6 degrees of freedom (for the definition of location and direction) are dispensed with: Sensor systems of this provenance already supply a reference coordinate system. Any such positioning may be sufficient in some fields of ultrasound (e.g., cardiology, prenatal medicine) - in others (e.g., endocrinology, dermatology) further adjustment may be necessary - therefore, such positioning is referred to herein as "primary positioning."

In any case, scanned-surface sonic models are an essential aid to correct positioning: as the transcutaneous ultrasound volume in the captured volumes captures the skin as a boundary "plane" and, as such, provides a geometric "surface fit" for positioning beneath the surface of the skin Sonic model, this information can provide the database for "fine-tuning algorithms" after a "primary Positioning "This said skin or boundary" plane "may be somewhat curved in the individual volume, but this method basically works as a sufficiently accurate approximation:

For the determination of "interfaces" already known methods for this from image processing (for example endocardial detection) such as grayscale gradient evaluation and the like can be used, with the last corrections possibly being made as "manual" rotations and translations.

This "post-processing" has already been addressed in Chapter IVS-DVA.

It should be concluded that the above description initially focuses on the "non-invasive" ultrasound (= transcutaneous, ie sounding through the superficial skin) For "invasive" ultrasound procedures such as TEE, IVUS, endosonography or the like, the method would have to be analogous to a Ortskoordinaten- table of the relevant internal (hollow or other) organs in the simulator system or for its corresponding invasive sound models are expanded, which in principle-technically makes only a small difference or the method is almost immediately transferable and subsequently neglected becomes. Such a location coordinate table, as in transcutaneous ultrasound, can be created with 3D sensors and further processed, be it magnetic field-based, inertial, optical or other 3D sensors. For the "fine tuning" explained in the last paragraph, other border structures such as esophagus, vascular, stomach, intestinal or other internal body wall structures can generally be used as an alternative to the skin.

6 VTP - Virtual Transducβr Processing

The PVP method (see above), in turn, is closely related to VTP, which detects the movements of the "virtual transducer", that is the transducer imitat, and evaluates it for simulation.

In particular, VTP determines the parameters explained below, which can be determined / derived from the surface scan data of the sound models, a geometric data description, in particular convex transducer imitations and from the current position data of the 3D sensor receiver in the transducer fidelity and are used for visualization. tion or simulation can be used. The scanned sound model surfaces provide a 3D dataset - assuming a static / fixed dataset at first - maW: If one neglects first the "indentation depth" with passive sound models (which is not necessarily - one can eg such caused organ displacements - as with a visualization of a liver - to a certain extent also with a passive sound model) - so at least haptic / active sound models also assume a dynamic data set of 3-dimensional spatial coordinates - quasi 4dimensional, however, in practice, with relatively few "representative states" temporal changes (eg maximum expiration - breathing medium - max inspiration or the like).

In the case of the 3D sensor technology of the transducer imitator, which in principle operates "freely in space", the necessity of a 6-dimensional system is to be assumed, for example for magnetic-field-based sensor systems: When acquiring the location data, the "magnetic field-generating" transmitter can be positioned as long as desired it remains in the working or accuracy range for the receiver attached to the transducer - this defines a reference 3D coordinate system within which the (typically dynamic) position of the receiver, which in turn is defined as a 3D data set, is determined - as a β-dimensional system. Sensor systems with a fixed reference coordinate transmitter, as e.g. using laser or camera-mounted 3D system components mounted on an ultrasound imaging stage - or equivalent components attached to the visualization / simulation sound model - or, ideally, both (also dynamic) SD spatial coordinates Through their constant or fixed reference coordinate system, they also operate quasi with a "calibrated" transformation of recording and subsequent visualization / simulation. Such 3D sensor components could e.g. be realized in the form of cameras with camera tripods or laser coponents. This initially refers to the 3D data sets of the (moving) acquisition transducer and the (moving) transducer imitation; in the said "ideal case", an "automatable" positioning of the (meta) volumes, s.a. Chapter (5) SSP, done - see also Chapter (10).

But first, for example, magnetic field-based 3D sensors with 6 degrees of freedom. Today's ultrasound simulators, by not taking into account the position of volumes relative to the imitated patient's body, the sound model, have some systemic weaknesses. Translational movements of the transducer compound are visualized by the simulator as a corresponding shift of the sectional image. By contrast, tilting or rotation movements can lead to unrealistically acting rotations of the volumes. These deficiencies are overcome with SSP and VTP because of the positioning of volumes relative to the sound model surface and the movement of the transducer imitative can be determined relative thereto. In addition, VTP allows the targeted visualization of desired additional, eg adjacent volumes, and in principle also has other possibilities, such as the implementation of a "look ahead" technique, in which the direction of movements of the transducer fiction is detected / analyzed and " Preparation "to be loaded volumes can be used. The parameters that can be determined by linking such sound model surface data and transducer-fidelity data and used for the visualization / simulation are in particular:

Center point C (= center point) of the placement surface of the transducer fiction.

In this case, C is actually the 3D sensor receiver position with a variable x, y, z offset depending on the type of transducer imitation used, which simulates a "linear" or "convex" transducer - and / or a different 3D sensor Receivers types.

The 3D sensor receiver is located within the transducer mimic at a defined distance from the "footprint" - theoretically this "touchdown point" can be considered as a mathematical point, i. infinitely small, but the replica of a transducer's sonic surface is not infinitely small like a mathematical point, but a surface (whose size and localization are also affected by touchdown pressure and direction). In practice, however, the simple determination of the geometry of sensor, sound model surface (see Chapter 5 "SSP") and the

Transducerlmitat form (s.u. GT) - immediately resulting "mathematical" point C suffice.

It should be noted, however, that in the case of convex transducer simulations in particular, the orientation in the center point calculation has to be taken into account: these transducer types are known for their circle-round sound surface "Tilting" (and to a certain extent also in the plane of the cut) predestined - and are handled as well in real as in the "virtual or simulation-sounding".

A geometric description GT (= Geometrical TransducerData) of transducers, especially those with convex sound surfaces. For the cross-sectional imaging typically used in the simulation, it should again be pointed out here that the "sound surface" of a transducer or transducer imitation is an area with a length and a width, but the piezo-crystals of a transducer of a 2D transducer on this surface are arranged side by side, so one (in linear transducers a straight line and) in convex transducers represent a "curved line" and thus produce a 2-dimensional sectional image. For such transducers, the geometric description of the circular section running through C with indication of the "radiation angle" and the radius is sufficient, only for 3D / 4D transducers, where the sound surface is actually a 2-dimensional image generator and thus generates a 3D / 4D image In the case of convex types, their imitates would have to be analogously given a "center" or mirror - symmetric "C - section" with "radiating range" and radius.

Axis CA (= C-Axis) and the sectional image plane CS (= C-Section-plane) through the point C: The "average sound beam" in a transducer - and thus also the "imaginary mean sound beam" of a transducer-imitative - is located in a defined or constant position - describable as the "centered" longitudinal axis of the transducer or transducer imitation.

C (see above) is defined as the last point of this longitudinal axis within the transducer and as the first point on the skin or the sound model surface along the "sound ray." This longitudinal axis defines the middle "sound ray" for all symmetrically built transducers real transducers as well as "imaginary mean sound beam" in transducer imitations.

This longitudinal axis is located in the middle of the real and the "virtual" (ie simulation) image intersecting plane CS For the geometric description in a transducing imitat CS is represented by CA (see above) and the "central" transverse axis CO (FIG. C-orthogonal-axis) of the transducer-imitative, which also intersects all points of the axis CA - thus the "virtual image slice" level of the transducer fiction by the two lines / axes CS and CO clearly described.

In the case of convex transducers, however, this description only applies in the case of an orthogonal exploration in the simulation - tilting a convex transducer approximately along the plane of the section moves the continent of the curved plane

Transverse axis to the left or right according to the rotational movement, which is considered position-dependent.

Polygon plane SP (= Spatial Plane of Polygonal Borderlines / Area) of the respective sound imager surface (defined by the point C described here above) (as X ₁ Y ₁ Z coordinate set). The scan method for sound models described in the previous chapter 5 "SSP" generates a large number of individual, coherent polygon planes SP, for example in the form of a table. Such a table contains the data of the boundary lines of these polygons as well as their position as an area in space. C (see above) is thus always exactly defined in one or on one of these polygons when the imitation of the transducer is attached. The polygon touched by C thus defines the plane of the "skin" or surface of the sound model in space.

On future possibly usable optical 3D sensors, this method is largely analog transferable.

The parameters described so far provide the necessary prerequisites for providing a geometrically realistic visualization / simulation of sectional images by recorded ultrasound (meta) volumes. For a further perfected visualization, a realistic localization / positioning of recorded (metric) volumes in the sound models is necessary:

For sound model localization / positioning, at least one slice image CS of a recorded (typically unpositioned) volume is now defined / stored as a "reference slice." Thus, the volume of the simulator system can be unambiguously adjusted to a desired spatial position Sound model - "pushed and / or rotated" by the stored section location data at a desired location for the visualization / simulation of the headgear imitation as reference data for a 3-dimensional position of the volume - thus provided with a translation and rotation - "rendered".

Typically, a user (doctor), who has a recording system and a "master" ultrasound simulator for building a case database with such

Volumes "feeds" or operates / operates the above described positioning first anatomically correct, but not so exactly that this volume now also optimally fits optimally under the surface of the sound model "pre-positioned" volumes often require a "post-correction" - in such a way that the volume touches as much as possible all geometrically affected polygon planes SP of the sound model, but nowhere protrudes beyond the same "outwards". Here VTP carries out a computational rework of the manual pre-positioning without further manual user interaction. Technically, the system could be used e.g. either the center of gravity of the axis or the center of the axis defined by C and the end point of the distance CA can be used as a fulcrum for rotations, and for translational motions the axis CA of the chosen "reference cut" could be used, for translations one could also use the average of two referential cuts - but other iterative or other methods are also conceivable for such an optimization method, the choice of which should be reserved for a method implementation.

Distance C-SP: The Transducer-Imitat should generate an image only if it is placed on the surface of the sound model or pressed into the sound model. Therefore, the distance in which the point C of the transducer fiducial is "above" the sound model or SP is continuously determined - these are positive values, for positive values the image visualization is switched off, starting from the value zero or for "negative "Values, however, the image is visualized. In addition, a shift of the skin level when pressing the

Transducer imitations in passive as well as in active / haptic sound models with the parameters listed in this chapter are simulated mathematically or visualizing. Reference sound axis RS: The possibility of perfecting this post-correction not only on the basis of the recorded volume limits themselves, but also on an interface determination and the skin level definition based thereon, has already been explained in Section 5 "SSP", with the exception of such a calculation The surface of the skin itself could also be calculated for this

Skin plane orthogonal scanning axis of the receiving transducer - e.g. as a "reference axis" for determining an "exploration delta" of current / current explorations. But it probably makes more sense to use the respective orthogonal axes, which in practice are hardly deviating from them, on the polygonal planes SP of the sound model.

(Conveniently, an orthogonal reference axis may also be simply stored with a corresponding prediction estimated as orthogonal to the user - i.e. the acquiring physician.) Thus, e.g. allowed limits of the Anlotung etc. are determined.

The parameters described so far relate to the optimal localization / positioning or sectional representation of (meta) volumes in the sound model. The acquisition and evaluation of the following further parameters also aims at an optimum performance of the system or computing speed.

Vector CV of point C during movement of the transducer fiction

Angular velocity and direction D-CA of the exploratory dynamics of the transducer fiction

By capturing the transducer-imitative motion as a vector with the direction and velocity of the touchdown point C - and capturing the dynamics of its exploration, a visualization / simulation system is also able to implement a look-ahead logic that can be used to calculate can be used, which volumes or meta-volumes are likely to be needed next. "Look-ahead" techniques are already used today in semiconductor memories of processors.

Possibly not all or even other parameters of this kind will be needed in praxi. The decisive factor is that the transducer imitat in its own geometry and the geometry not only in relation to the (meta) volumes, but also is recorded and evaluated with reference to the sound model or its surface or as a relationship of (meta) volume, sound model surface and transducer-fidelity data.

Of course, the described "tracking" of the transducer imitation can also be stored - and thus also in every detail the sound head guidance of a trainee on the simulator for subsequent evaluations, etc.

7 AHS - Active + Haptic Scan Models

Depending on the field of application (gynecology, gastro-enterology, etc.), sound models (patient phantoms, also referred to as "mannequin" or "doll") are passively sufficient - or not. So far, at least, only passive sound models have been used. "Passive" means that a sound model only represents a 3dimensionale surface for the putting on of a "Transducer-Imitats" or "Pseudo-Transducers" of an ultrasonic simulator, but otherwise has no further own functionality or can supply technical information - in the In the field of gynecology, especially in the field of prenatal medicine, such passive sonic models have proved to be adequate so far, in which breathing plays practically no role, and in the case of sounding only little cooperation between the patient is necessary - and accordingly there is a simulation for the patient herself or as a patient's own functionality practically nothing to emulate, apart from the fact that it can be sounded.Such sound models are therefore made simply from plastic or foam.

Apart from taking a half-lateral position, echocardiograhy also involves respiratory organ shifts (lung / heart), and especially for important parts of abdominal sonography, this is even more true - also as a deliberate means of visualizing some organs - playing here So the patient's interaction also plays an important role in imaging.

In gastro-enterology, the "depth of impression" and similar haptic parameters are sometimes of considerable importance as they stand today and are desirable for the realism of ultrasound simulation. len organs is often a prerequisite for their visualizability. This is achieved by targeted impressions of the transducer and breathing maneuvers with diaphragmatic breathing. In emergency medicine, for example, the reaction - a pain-related defensive tension of the abdominal muscles - of the patient also plays an important diagnostic role.

For this purpose, a sound model with functions such as patient respiration, reaction to the touchdown pressure of the transducer (via manual transducer guidance of the "virtually scanning" trainee) is used, for example, with a pneumatic system whose states can be controlled with a microprocessor. Thus, a position, pressure and direction-dependent, organ-specific visualization including an inclusion of haptic (= sense of touch-oriented) functions in the sound models or corresponding training option on the simulator allows.

A realization is not necessarily bound to pneumatics - only the ability, a suitable for simulation to achieve certain conditions such as patient on / off breathing - possibly controllable - variable surface, and / or the elastance or to generate compliance behavior of the sound model surface. This should be given for defined states. This could possibly also be realized by appropriate surface materials, by underlying springs, etc. A pneumatics with controller control is used here only as an obvious example of a process implementation, because thus also an active control of desired states and a data / control communication with the image processing computer is well represented.

Such sound models are described here as "active sound models" - in the current development phase these are also "haptic sound models", but this is not necessarily the case - an active but not haptic sound model (eg only respiration) is just as conceivable as a haptic (eg with underlying corset skeleton and spring pressure working), but not active sound model. If necessary, this "active / haptic sound model" system will be described separately in the technical description of the simulator, with different parameters or scenarios of spatial position and pressure of the transducers corresponding to different ones for the later simulation Recordings (typically similar sound structure such as an organ) taken into account.

During the "rendering" of the (meta) volumes, both the (to be preselected by the software) patient status and the interactive (manually generated by the "trainee") transducer status are now simulated by the simulator system for a realistic simulation If necessary, an appropriate reaction of the patient model is carried out on actions of the user / trainee, and a (meta) volume relevant for the respective scenario is visualized.Chapters 2 to 6 are content-related or mathematical , Algorithmic and technical approaches are partly interconnected, so the implementation of this volume visualization will be explained in more detail in the following chapter.

The recording and visualization of the "fitting" (meta) volumes for the different states of such sound models and their interaction with the transducer imitation correspond to the principle of the PVP method.

8 ADP - Angulation-Related Doppler Processing

Doppler ultrasound has a fundamental difference to the usual gray value images of ultrasound such as B-mode or analogue visualizations with 3D / 4D scanners: the morphology image changes with the transducer movements, but not in their geometrical or perspective In terms of content, in Doppler, the movement of the transducers also means the (often visualized by color coding) meaning of the parametric image components - ie the blood flow shown in Doppler - indirectly changed:

Red typically indicates a flow towards the transducer, blue away from the transducer. If a vessel is sounded, for example, with a bevelled position opposite to the direction of flow - and then tapered in the same direction as the flow direction, these colors will be reversed in the image, although the same vessel is sounded over the same anatomical site, but in two different approaches. If one now acquires a vessel with Doppler technique - via sweep with constant sounding as in gray-scale image recordings - for the later visualization / simulation, one would have to at least pretend to the trainee in which possible direction of sounding he should sound in the simulation. Obviously, this is an absolutely undesirable limitation, because the ultrasonic simulator is supposed to teach you the right soldering - and not be used as a working basis.

As already explained, for ultrasound simulation it is first necessary to work with a recording and then with a reproduction (simulation) system - in the case of Doppler, tuning both systems becomes particularly important.

For peripheral angiology / phlebology, to solve the problem described here, the 3D sensor attached to the recording transducer can also be used to record the acquisition during acquisition - the software provides features in accordance with the following description of an example solution ,

Recording: The image / volume-receiving physician should first save the othogonal sound of a vessel to be detected - then the course of the vessel or vascular segment to be recorded: This is basically the initial calibration of the following image. This axis defines the center of an imaginary cone whose base is thus approximately parallel to the vessel - and thus orthogonal to the defined reference Anlotungsachse. The cone base is in turn divided into 2 mutually orthogonal lines - one of which is roughly in the defined direction of the vessel to be detected and the other "transverse" - divided into four equal segments. The doctor now picks up the vessel in the middle of each segment of the circle, ie a total of 4 times, ie after said "calibration" along the course of the vessel in the 4 different approaches described above. To facilitate such recording, a small "LED traffic light" can be made available, indicating how well you are in the defined, selected sounding.

Playback: On the simulator, the technique described in Chap. 6 "VTP" captures the exploration of the transducer fiction and now visualizes - in accordance with the selected animation - the "right" volume with the "correct" color coding from the 4 available. ren for each vessel or each vessel section. Thus, approximatively, but sufficiently accurately (see introduction: Doppler and accuracy of the same) a real (eg color) Doppler can be simulated.

Refinements of this technique are of course possible over a larger number of said "conic sections" - but increases the cost of recording in proportion.Furthermore parameters (see Chapter 6 "VTP") as the vector CV of the point C during movement of the Transducer mimetics or the angular velocity and direction D-CA of the exploratory dynamics of the transducer fiction can be used to achieve a desired "inertia" in the visualization: a "dithering" into another conic section should not be for a fraction of a second retrieve other conic section, etc. Such considerations are more of a question for

Method implementations.

In the case of acquisition and reproduction, only "orthogonally from the top" leads to a usable result for the visualization / simulation of Doppler.The deciding factor is the basic consideration or its realization that for a simulation of the Doppler ultrasonography an exploration-related Recording takes place and the simulation realizes an analogue-related visualization.

This method is primarily intended for angiology / phlebology in Doppler, but can also be transferred analogously to cardiology (other possible uses see note in Chapter 3 "PVP").

9 SIA - Soft Tissue Immersion Meta-Volume Acquisition

Soft tissues without sufficiently supporting and / or associated solid structures are a significant problem for the detection of (meta) volumes for simulation. In most areas of the body, even in the relatively soft abdomen, the skin of "top" and surrounding structures inside the body such as other organs, bones, etc. limits the displacement of a structure to be recorded - only more specific pressure then leads to even the abdomen Thus, when performing a scan or sweep, a continuous / steady, gentle pressure with the imaging transducer during a sweep can provide a sufficiently stable / stable imaging situation that allows frames to be joined together (at 2D => 3D). Conversion) or a joining together of volumes allowed. Although the abdomen of a very obese patient or especially a female (especially larger) breast can be sounded, it does not allow the creation of the above-described necessary recording conditions - frames or volumes of any such recordings would no longer make it possible to combine them. Thus, only the reconstruction or visualization / simulation of frames or (when used for recording volume scanner) of individual, small volumes, but no longer larger areas or even an entire breast directly achievable.

The classic and most important example of this problem is now exactly one female breast in mamma sonography. As a solution to such problems, a scannable container with "sound flow path", a so-called "water bath" is often proposed. Such a water bath construct fulfills the goal of generating a constant shape during image acquisition: Gravity keeps the breast immersed in the water from above in a constant shape during the scanning process. However, this does not solve the problem that such soft tissue without support structures such as a mamma may be individually "highly different" shaped. the outer shape of a volume of considerable importance, because this external shape of a volume must be fitted into the sound model. However, volumes of individually different soft tissue structures and shapes, such as the highly individualized female breasts sensed by a water bath, may not typically be "appropriately" projected into the uniformly shaped breast of a sound model. The "reverse" way of equipping a sound model with as many interchangeable breast forms as there are varying "shapes" is very cumbersome and hardly practicable for courses or trainees.

Therefore, an immersion method was conceived and realized here as well as a first experimental design that generates the desired outer anatomical shape by means of approximately half-shell-shaped immersion containers whose at least front part is shaped exactly like an impression of the sound model breast. and having the properties of a water bath by the choice of a material having water-like acoustic properties. More specifically, these were "breast baskets" that were installed in various sizes as approximately tapered shells with uniform pitch (interchangeable) in a prone position couch. For the acoustic coupling of the transducer, these shells can be filled with lukewarm water before immersing the breast or coated inside with the otherwise used ultrasound gel - on the outside of the sound is also brought on ultrasound gel on. The recording physician then sounds through these forms from below on a pedestal in sitting height, on which this couch is also lying with the patient in prone position.

Very large breasts, which are so captured, then protrude slightly further than anatomically true in the following simulation in the area, which would actually represent the thorax, but this does (the thorax inside is not soundable because of the air filling anyway) The purpose of the visualization / simulation virtually no demolition, as medical experts have confirmed.

This immersion method can be used both for volume scanners and for the conventional / previous method of acquiring frames with 2D scanners, which are then converted into a volume. Initial tests on such a recording couch with such breast inserts for manual scans were promising and are a novelty.

Instead of shells having a conical shape on the inside (a slightly smaller one) and on the outside (because of the wall thickness of the shell material a little larger), variants are also possible which have a conical shape inside and outside, e.g. In this case, the differently sized cones for the different breast sizes would be "milled out" into cylindrical blocks made of this material, which would make it relatively easy to realize a motor for attaching a recording transducer - ie this method is otherwise also automatable and would then introduce a prerequisite for the standardization of breast somography, which is also purely diagnostically desirable or often required (the lack of standardization is seen in today's breast ultrasound as the most significant deficiency. Compared to the radiological mammography ).

For this method theoretically also with the conventional 2D => 3D method good results are to be expected, because the usual disturbing factors (respiratory / movement artifacts etc.) can all be turned off quite elegantly - and an almost arbitrarily slow and even motorization The scan process implies a good geometric resolution per se. But there are still no practical a good geometric resolution. But there are still no practical tests / results available.

An automation with volume scanners is also conceivable; It should, however, be clarified whether the effort of motorization / automation pays off here, because the scanning itself should already be sufficiently simple and qualitative by the methodology (see especially Chapter 2 IVS - DVA) per se and should be reproducible / uniform.

9 Future Ultrasound Acquisition + Visualization / Simulation Technologies

The examples presented are based on technology available today. However, a number of HUMR processes already require technologies that have only been available since 2006. For example, The DVA method, which reads volumes in the simulation more or less directly from the hard disk, can scarcely be handled simultaneously with a sound model-related tracking of the transducer fiction, as provided by the VTP and method. Both high transfer speeds for disks and multi-core processors or multi-threading of the corresponding programs should be prerequisite for the feasibility, as long as the real-time behavior of the simulation achieved today should be maintained. By contrast, the alternative IVS method and PVP, when applied extensively, could put some demands on the RAM capacities.

In this respect, HUMR is so z.T. relatively precisely matched to the technical possibilities - which, however, will continue to change rapidly. So here is the goal to see clearly - and that means a "realistic visualization / simulation." Extrapolating the development of ultrasound in the last 20 years to the next 10 years, HUMR is likely at a hypothetical completion time in 2 years for This period - about a decade - will be a good solution for ultrasonic simulation.

Further changes can be expected in the Internet sector - in particular a significant improvement in bandwidth, be it cable or wireless. Thus, in the future, downloads of casuistry, ie the case database modules, will play a major role in the simulation. There may be a need for supportive online tutoring and the like. Like. arise. In the future, the need for training will tend to be more in the direction of a broad and basic education than in the direction of further education for advanced students, although it will continue to do so. Paradoxically, just the basic and basic training requires a further perfection of the simulation, because beginners and the less experienced are more irritated by limitations of the simulation than the advanced. On the other hand, this will also increase the demand for cheaper simulators.

The most expensive hardware component of the simulation (to be distinguished from the detection technology) today is 3D sensor technology. Here, optical sensors will play an important role, since they can be produced significantly cheaper. Here, the simulation is still focused on the current magnetic field-based 3D sensor technologies, but this patent application expressly formulated this as an example and points to the transferability or points to the need to have to adapt some procedures on it, even the basics of Use of Optical SD Sensors with Fixed Spatial Reference Coordinate System for Acquisition and Visualization /

Simulation were described. This is finally pointed out again.

Claims

claims

A method of processing ultrasound image volumes wherein SD ultrasound image volumes are acquired using a transducer of a 3D ultrasound scanner, characterized in that position and, during detection of the 3D ultrasound image volumes Detected orientation of the transducer and associated with each detected SD ultrasound image volumes, captured 3D ultrasound image volumes under evaluation of the respective position and orientation with other recorded 3D ultrasound image volumes to an extended ultrasound image Volume (meta-volume) are assembled and stored, and / or that the recorded 3D ultrasound image volumes are stored separately with the linked data on position and orientation.

2. The method according to claim 1, characterized in that the processing comprises a restitution of ultrasound image volumes, wherein for restitution of the ultrasound image volumes position and orientation of a

Detected sensor and depending on the detected position and orientation of the sensor, a partial volume of the extended ultrasound image volume is restituiert and visualized.

3. The method according to claim 1, characterized in that the processing comprises a restitution of ultrasound image volumes, wherein for the restitution of the ultrasound image volumes the position and orientation of a sensor are detected and recorded on the individually stored 3D ultrasound image. Volume is accessed, whose data on position and orientation correspond to the detected position and orientation of the sensor, and this 3D ultrasound image volume is visualized.

4. Method according to claim 2, characterized in that the sensor is a transducer-imitation and in the restitution an SD-ultrasound image volume is restituted which essentially has a 3D ultrasound image in its geometry. Volume that would be produced by the imitated transducer.

5. The method according to any one of claims 1 to 4, characterized in that the restitution takes place in the context of a simulation of an ultrasound examination.

6. The method according to any one of claims 1 to 5, characterized in that for positions and orientations over a period of time in each case a plurality of SD

Ultrasonic image volumes are recorded to detect temporal, especially periodic, changes in sounded areas.

7. The method according to claim 6, characterized in that for at least part of the positions and orientations, the plurality of 3D ultrasound image volumes acquired over a period of time are combined for each position and orientation into 4D ultrasound image volumes.

8. The method according to any one of claims 1 to 7, characterized in that a parameterization of the 3D ultrasound image volumes is carried out.

9. The method according to any one of claims 1 to 8, characterized in that for generating the expanded ultrasound image volume from a first SD ultrasound image volume (start volume) each generated by changing the position and / or orientation of the transducer Next SD ultrasound image volume without the overlap area is added to the startup volume, and this enlarged 3D ultrasound image volume progressively Movement of the transducer, in turn, as the seed volume is increased by adding the next volume, etc., or that the new 3D ultrasound image volume to be added overwrites the already existing 3D ultrasound image volume over the overlap area in the previously detected volume.

10. The method according to any one of claims 1 to 9, characterized in that

Surface position data of sound models is scanned and taken into account in a simulation of ultrasound examinations.

11. The method according to claim 10, characterized in that derived from scanned surface position data of sound models position and motion data of a transducer fiction.

12. The method according to claim 10 or 11, characterized in that are described by the surface position data motion states of sound models.

13. The method according to claim 12, characterized in that it is a haptic sound model in the sound model.

14. An arrangement for processing ultrasound image volumes, the arrangement comprising a 3D ultrasound scanner with a transducer, a data processing device and storage means, characterized in that the arrangement further comprises a sensor for detecting the position and orientation of the transducer and is arranged such that a method for processing ultrasound image volumes is feasible, wherein 3D ultrasound image volumes are detected using the transducer, During the acquisition of the 3D ultrasound image volumes, the position and orientation of the transducer are recorded and linked with the respectively acquired SD ultrasound image volumes, 3D ultrasound image volumes acquired by evaluating the respective position and orientation with other detected ones 3D ultrasound image volumes are merged and stored into an expanded ultrasound image volume (meta-volume), and / or the captured 3D ultrasound image volumes are individually stored together with the associated positional and orientation data ,

15. Arrangement according to claim 14, characterized in that the arrangement further comprises a transducer-imitat and is arranged such that a restitution of ultrasound image volumes can be carried out, wherein the restitution of the ultrasound image volumes position and orientation of the trans - Ducer-Lmitats detected and depending on the detected position and orientation of the transducer fiction, a partial volume of the extended ultrasound image volume is restituiert and visualized.

16. Arrangement according to claim 14, characterized in that the arrangement further comprises a transducer-imitation and is arranged such that a restitution of ultrasound image volumes can be carried out, wherein the restitution of the ultrasound image volumes position and orientation of the trans - ducer-lmitats and recorded on the individually stored SD-

Ultrasound image volume is accessed, whose data on position and orientation correspond to the detected position and orientation of the transducer fiction, and this 3D ultrasound image volume is visualized.

17. Arrangement according to one of claims 14 to 16, characterized in that the arrangement further comprises a sound model and is arranged such that a simulation of an ultrasound examination is feasible.

18. Arrangement according to claim 17, characterized in that it is a haptic sound model in the sound model.

19. A computer program that allows a data processing device, after being loaded into storage means of the data processing device, to perform a method for processing ultrasound image volumes, wherein a 3D ultrasound scanner, a transducer for acquiring 3D ultrasound image Volume and a sensor for detecting the position and orientation of the transducer via means for data transmission to the data processing device are connected, and wherein detected during the detection of the 3D ultrasound image volumes position and orientation of the transducer and with each detected SD ultrasound Image volume, captured 3D ultrasound image volumes by evaluating the respective position and orientation with other acquired 3D ultrasound image volumes to form an expanded ultrasound image volume (meta-volume) and stored, and / or that the captured 3D ultrasound image volumes are too stored separately with the linked data via position and orientation.

20. A computer-readable storage medium having stored thereon a program that allows a data processing device, after having been loaded into storage means of the data processing device, to perform a method for processing ultrasound image volumes, wherein an SD

Ultrasonic scanner, a transducer for capturing SD ultrasound image volumes and a sensor for detecting position and orientation of the transducer are connected via data transfer means to the data processing device, and wherein during the detection of the 3D ultrasound image volumes Detected position and orientation of the transducer and are linked to each detected SD ultrasound image volumes, captured 3D ultrasound image volumes under evaluation of the respective position and orientation with other recorded 3D ultrasound image volumes to one expanded ultrasound image volume (meta-volume) and stored, and / or that the acquired 3D ultrasound image volumes are stored separately with the associated data on position and orientation.