WO1988003275A1

WO1988003275A1 - Method for reconstructing images of successive parallel section cuttings of an object containing gamma radiation emitting sources

Info

Publication number: WO1988003275A1
Application number: PCT/FR1987/000421
Authority: WO
Inventors: Alain Lansiart
Original assignee: Commissariat A L'energie Atomique
Priority date: 1986-10-31
Filing date: 1987-10-27
Publication date: 1988-05-05
Also published as: EP0329685A1; JPH02500539A; FR2606160A1; FR2606160B1

Abstract

Method for reconstructing images of successive section cuttings of an object (2) by means of a gamma-camera (5) having a plane detector (1) parallel to the section cutting planes. The method comprises the use of a mask (M) arranged between the object and the camera, forming a coding matrix comprising patterns (M1) formed by transparent elements in an opaque surface. The method comprises then the processing of signals provided by the gamma-camera, from a matrix for decoding the images of the sources on the detector, obtained by traversing the mask. The processing uses an iterative algorithm of signal processing. Application to the formation of tridimensional images of objects.

Description

METHOD FOR RECONSTRUCTING SUCCESSIVE PARALLEL CUT IMAGES OF AN OBJECT CONTAINING GAMMA RADIATION SOURCES

DESCRIPTION

The present invention relates to a method of reconstructing images of successive parallel sections of an object containing sources of gamma radiation, using a gamma camera with a plane detector parallel to the section planes. It applies to the reconstruction of images of parallel sections of an object, and in particular to the reconstruction of tomographic images of one or more successive parallel sections of an organ of the human body, containing sources emitting radiation gamma. We know that it is possible to reconstruct the images of successive parallel sections of an object or an organ containing sources emitting gamma radiation, using a gamma-camera with fixed plane detector, associated with a plane mask, parallel to the detector; this mask comprises an assembly of patterns each formed by a coded mosaic of transparent elements opaque to the radiation produced by the sources contained in the object or the organ.

The basic principle used in obtaining an image from a coded mosaic of transparent and opaque elements is simple. A point source of radiation causes the appearance of an image of each transparent element on the detector. The size of the image depends on the distance from the source of the detector, while the location of the image on the detector depends on the lateral displacement of the source relative to the detector. If each pattern of the mask comprises N transparent elements, one obtains on the detector N images of a source occupying predetermined positions in a fixed reference frame of reference. The positions of the N transparent elements of the mask are coded; it follows that for a source having a predetermined position in a plane parallel to that of the detector and for a predetermined distance from this plane relative to the detector, N images corresponding respectively to these N transparent elements are formed on the detector, in coded locations depending on the position from the source in his plan. The size of each image depends on the distance of the plane containing the source from the detector. The coded information concerning the coded positions of the images, the dimensions of these images, as well as the intensities of the radiation received by the detector at the locations of these images are recorded in a memory.

This recorded information does not allow direct reconstruction of the images of different sections of an object, in planes parallel to the detector. Indeed, the images of the transparent elements of the mask, obtained by means of radiation from different sources of the same plane or from different planes, overlap. Correct images of. cuts of an object or an organ, can only be obtained thanks to reconstruction methods which make it possible to determine the location of each source, as well as the intensity of the radiation emitted by it; these methods use processing of the information stored in memory, which are of two types: processing by deconvolution or processing by correlation. 5 -. The principle of reconstruction of section images of an object using a gamma-camera with plane detector, associated with a mask with transparent and opaque elements, by convolution or correlation processing of the signals supplied by this gamma -camera, is described in an article (1) entitled "Coded _Q aperture imaging with unifor ly redendant arrays" by EEFENIMORE and TM CANNON- review "Applied Optics" - Volume 13, n ° 3, February 1, 1978, pages 337 to 347.

Another article (2) by the same authors, entitled

"To ographical imaging using uniformly redendant arrays" in the ₅ review "Applied Optics" - Volume 18, n ° 7, Apπ ^' l 1, 1979, pages 1052 to 1057 describes more particularly the application to tomography, of the principle of the preceding article (1). Another article (3) entitled "An evaluation of techniques for stationary coded a aperture three-dimensional imaging in nuclear medicine" 5 by JS FLEMING and BAGODDARD, published in the journal "Nuclear instruments and methods in physics research" 221-1984- North Holland, pages 242 to 246, describes a method of reconstructing sectional images of an object or an organ by processing deconvolution of information from the detector of a gamma-

10 camera associated with a mask.

In general, the treatments carried out according to the known methods are long and complicated and an unsolved problem, whatever the treatment method used, is the significant background noise which appears in the reconstruction of

15 each image; the increase in the ratio of the signal (useful for the reconstruction of images), to the noise (notably created by the sources surrounding that which provides a useful signal), is not easy to obtain. The first article (1) shows in particular that a deconvolution treatment provides a bad 0 reconstruction of images because noise remains predominant. Better image reconstruction is obtained by correlation processing which significantly reduces noise. This is moreover confirmed in the two other articles (2), (3).

The improvement of the signal / noise ratio can be

25 obtained, as shown in particular in articles (1) and (2) by arranging the transparent elements of a mask according to a random distribution. This distribution which improves the value of the signal / noise ratio does not however make it possible to obtain extremely satisfactory images, the processing of the information

2Q by correlation failing to significantly increase the value of the signal / noise ratio. In these known devices, the transparent elements are numerous and occupy fifty percent of the surface of the mask.

The object of the present invention is to remedy the ^"

“Disadvantages of known methods of reconstructing images of sections of an object, and in particular to provide a rapid process making it possible to very significantly reduce the effect of background noise on the reconstructed images. This object is achieved, in particular by using a mask in which the transparent elements of each pattern are distributed in a predetermined pseudo-random manner, and by carrying out a particular iterative processing of the count values supplied by the gamma-camera.

The invention relates first of all to a method of reconstructing images of successive parallel sections of an object containing sources emitting gamma radiation, using a gamma-camera with plane detector parallel to the section planes, this camera having a plurality of outputs each identified by a code and corresponding respectively to the different possible impact coordinates and to the energy of the gamma rays received by the detector, this energy is measured by a digital value supplied on the output corresponding to a impact, the method consisting in interposing between the object and the detector, parallel to this detector, a plane mask formed by assembling identical patterns, of rectangular shape and each comprising a mosaic of transparent elements in a surface opaque to radiation, these transparent elements occupying in the pattern predetermined positions identified according to a coding matrix, the reconstruction of the images of the sections being carried out by processing the count values recorded in a memory, these values being selected according to decoding matrices deduced from the coding matrices and corresponding to the images on the detector of at least one pattern of the mask illuminated by said sources, the dimensions and the locations of these images on the detector depending on the pattern chosen in the mask, the coordinates of the sources in each plane, with respect to a reference frame, as well as the distance of each plane from the detector , the method consisting in carrying out the following operations: a portion of the detector is selected for each distance from the cutting plane Pi relative to the detector,

- for each cutting plane Pi, the decoding matrix is moved to locate the sources possibly present in this plane, this displacement corresponding to a scanning of all the sources identified by their coordinates (i, j) in the plane Pi, and by measured count values, the sources identified being those located on lines passing through the geometric center of the detector, - from the corresponding count values, the activities of the sources of each plane are reconstructed, characterized in that the the following treatments are then carried out:

- for each predetermined distance plane Pi, a comparison is made between the activity of each source of coordinates

(i, j) thus reconstructed in the plane Pi, and the respective reconstructed activities for neighboring sources located in neighboring planes Pi + 1 and Pi-1, having the same coordinates (i, j) as the source considered in the plane Pi, - it is determined whether the differences between the activity of each source of coordinates (i, j) in each plane Pi and the respective activities of the sources of same coordinates (i, j) in the plans Pi — 1 and Pi + 1 neighbors exceed a predetermined threshold, this exceeding indicating the location of a source of coordinates (i, j) in the plane Pi,

- an activity is assigned to each identified source,

- we calculate for each identified source the count value that this source would produce from the activity assigned to it, - we subtract, for each identified source, the count value actually measured previously for the source with the same coordinates ( i, j) in the same plane, of the count value calculated for the identified source, so as to obtain a corrected count value for the identified source, - the activity of each identified source is determined from La corresponding corrected count value and the previous treatments are repeated in an iterative mode until the corrected value obtained for each source is invariable. According to another characteristic, each of said transparent elements of a pattern is identified by a row number and a column number, the total number of rows L multiplied by the total number of columns C being equal to the length of a one-dimensional series. pseudo-random, L and C being prime to each other.

According to another mode of implementation. The method of the invention consists in reconstructing images of successive parallel sections of an object containing sources emitting gamma radiation, using a gamma-camera with plane detector parallel to the section planes, this camera having a plurality of outputs each identified by a code and corresponding respectively to the different possible impact coordinates and to the energy of the gamma radiation received by the detector, this energy is measured by a digital value supplied on the output corresponding to an impact, the process consisting ^" of interposing between the object and the detector, parallel to this detector, a plane mask formed by assembling identical patterns, of rectangular shape and each comprising a mosaic of transparent elements in a surface opaque to radiation, these transparent elements occupying in The pattern of predetermined positions identified according to a coding matrix, the reconstruction tution of the images of the sections being effected by processing the count values recorded in a memory, these values being selected according to decoding matrices deduced from the coding matrices and corresponding to the images on the detector of at least one pattern of the mask illuminated by said sources, the dimensions and locations of these images on the detector depending on the pattern chosen in The mask, the coordinates of the sources in each plane, with respect to a reference frame, as well as the distance from each plane by report to the detector. The process consisting in carrying out the following operations:

- a portion of the detector is selected for each distance from the cutting plane Pi relative to the detector, - for each cutting plane Pi, the decoding matrix is moved to identify the sources possibly present in this plane, this displacement corresponding to a scan of all the sources identified by their coordinates (i, j) in the plane Pi and by measured count values, ~ from the corresponding count values, the activities of the sources of each plane are reconstructed, characterized in that each of said transparent elements of a pattern is identified by a Line number and a column number, the total number of lines L multiplied by The total number of columns C being equal to the Length of a pseudo-random one-dimensional series, L and C being prime between them, the bidi ensional series A (i, j) being obtained from the one-dimensional series - A (x) by writing that the binary content of A (i, j) = A (x with i = mod x and j≈mod x, the series retained rs being q-nary and satisfying the equations: n n-1 n-2 q -1 q -1 q -1 i _m = -_. _p = --_. _; (. = -_._ _e q = q-1 q-1 q-1

k is a prime number and i is an integer, m represents the number of elements of a mask pattern, p the number of holes or transparent elements, l the correlation of the series by itself.

According to another embodiment, the method of the invention consists in reconstructing images of successive parallel sections of an object containing sources emitting gamma radiation, using a gamma-camera with plane parallel detector. in section shots, this camera having a plurality of outputs each identified by a code and corresponding respectively to the different coordinates of possible impacts and the energy of the gamma rays received by the detector, this energy is measured by a digital value supplied on the output corresponding to an impact, the process consisting in interposing between the object and the detector, in parallel with this detector, a plane mask formed by assembling identical patterns, of rectangular shape and each comprising a mosaic of transparent elements in a surface opaque to radiation, these transparent elements occupying in the pattern predetermined positions identified according to a coding matrix, the reconstruction images of the sections being carried out by processing the count values recorded in a memory, these values being selected according to decoding matrices deduced from the coding matrices and corresponding to the images on the detector of at least one pattern of the mask illuminated by said sources. The dimensions and locations of these images on the detector depend on the pattern chosen in the mask, the coordinates of the sources in each plane, relative to a reference frame, as well as the distance of each plane from the detector. The process consisting in carrying out the following operations:

- selecting ^a portion of the detector for each Pi cutting plane away from the detector,

- for each cutting plane Pi, the decoding matrix is moved to locate the sources possibly present in this plane, this displacement corresponding to a scanning of all the sources identified by their coordinates (i, j) in the plane Pi and by measured count values,

- from the corresponding count values, the activities of the sources of each plane are reconstructed, characterized in that it consists, for each cutting plane Pi, of selecting a detector portion corresponding to at least two projections of the same pattern of the mask on the detector, these projections being as far apart as possible.

The characteristics and advantages of the invention will emerge more clearly from the description which follows, given in reference to the attached drawings in which:

- Figure 1 is a diagram which makes it possible to better understand the use of a mask interposed between the camera and the object, in the method of the invention, - Figure 2 is a table making it possible to determine the distribution of the elements transparencies in each pattern of the mask used in the invention,

- Figure 3 shows schematically One of the mask patterns, ~ Figure 4 shows schematically a system for implementing the method of the invention.

FIG. 1 provides a better understanding of the principle of a method for reconstructing sectional images of an object, using a gamma camera associated with a mask comprising patterns with transparent elements as used for example in the method of CANNON and FENIMORE mentioned above, or in the process of the invention.

In this figure, the screen is shown schematically with a circular periphery of a detector 1, of a gamma-camera. Gamma radiation is emitted by sources located in an object or an organ 2. The detector 1 is a plane detector without collimator. The coordinates of the points of impact of the radiation on this detector are identified with respect to a reference frame X, Y, the origin of which is, for example, confused with the geometric center 0 of the detector. The sources belong for example to different section planes P1, P2, P3, of the organ or of the object 2, parallel to the plane P of the detector 1. The distances from these planes relative to the detector are marked on the axis 0Z perpendicular to the XY axes of the reference frame. The coordinates of the sources are identified with respect to the reference frame OXY, and by the respective distances Z1, Z2, Z3 from the planes P1, P2, P3 which contain these sources. These distances are identified on the axis 0Z, for example relative to the plane P of the detector 1. A mask M, contained in a plane P is interposed between the detector 1 and the object or the organ 2, of which we want to reconstruct the images of the sections, in the planes P1,

P2, P3 for example. The plane P is parallel to the plane of the

M detector. This mask comprises an assembly of patterns such as the pattern M1, each formed by a mosaic of transparent elements 3 and opaque 4 to the radiation emitted by the sources. To simplify the figure, only one motif of the assembly of motifs constituting the mask has been shown. Likewise, only a few transparent and opaque elements of the mosaic constituting each motif have been represented. Each pattern, such as M1 for example, illuminated by a source of gamma radiation, has an image on the detector 1. SeLon The assembly of the patterns in the mask M, the images of these patterns are organized according to a matrix having a well shaped defined; depending on the assembly of the mask patterns, and specific to each source. The extent of each image of a pattern depends on the distance from the source producing that image. The dimensions of the images of a pattern, as well as the intensity of the radiation received by the detector for each image, therefore depend on the coordinates of the corresponding source relative to the reference frame, and on the distance from the plane of the source by report to the detector. The plane dimension of the object is limited by the object-mask and mask-detector distances. On the other hand, with a mask made up of numerous basic patterns, it is possible to choose in the plane of the detector a portion of this detector corresponding to a projection of any pattern of the mask for a predetermined distance from the cutting plane containing sources in the object to be reconstructed.

If we take a source S1 in a plane P1 parallel to the plane of the detector, this source being located at the intersection of the axis 0Z passing through the center 0 of the detector and perpendicular to the plane P of this detector, the image on the detector of the pattern M1 of the mask M lit by the source S1 is represented at 11. The extent of this image represents the useful portion of the detector, used to reconstruct the images of the sources contained in the plane P1 located at a distance detector preset.

The maximum field of the section C1 of the object which it is possible to observe for the plane P1 is obtained by projecting from the center 0 of the detector, the same pattern M1 of the mask, on the plane P1.

For another source S2 of the plane P1, which is not located on the axis 0Z, the same portion 11 of the detector is used, by projecting onto it another pattern M2 of the mask M.

If a source S3 is located on the axis 0Z, but in a plane P2 further from the detector than the plane P1, the image 13 of the pattern M1 illuminated by this source S3 covers a more restricted portion of the detector, than the image 11 produced by the source

S1.

On the contrary, if a source is located on the axis 0Z, in a plane closer to the detector than the plane P1, the image of the pattern M1 illuminated by this source covers a more extended portion of the detector than the image 11 produced by Source S1.

If, starting from the center 0 of the detector, any straight line is drawn, cutting for example three consecutive cutting planes, very close together, respectively at three points, these points have substantially the same coordinates (i, j) with respect to the coordinate system (X, Y) of reference; only the distance z from these points marked on 0Z varies with respect to the detector. We seek the maximums of intensity of the sources, by counting, for sources which are located in parallel and neighboring section planes and which are on the same straight line passing through the geometric center of the detector, these sources having the same coordinates (i , j) in each plane.

Starting from the center 0 of the detector and relying on a mask pattern, as indicated above, an envelope of the object is defined. All the sources must be located in this envelope, but for each cutting plane, the detector portion is chosen so as to correspond to any pattern of the mask. A gamma camera equipped with such a type of detector and using a mask constituted by an assembly of patterns, provides on outputs, numerical values relating to the energies of gamma radiation received by the detector, through the transparent elements of the mask. Each output is coded in position and therefore also makes it possible to locate the coordinates, relative to a reference frame, of the impact of radiation on the detector. The numerical values for each impact position can be recorded before undergoing a processing allowing the reconstruction of the images of the sources of each cutting plane. The values processed are those which correspond to the images of the patterns, for each distance from the cutting plane and for all the coordinates of emitting sources which may be located in each plane.

In a known manner, according to CANNON and FENIMORE, we decode for all the section planes and for all the coordinates of emitting sources which may be located in these planes, all the count values corresponding to all the images of all the patterns of the mask. This decoding is carried out using matrices for decoding the images of the patterns on the detector. These decoding matrices are obtained from the coding matrices. This decoding makes it possible to select from the count values, those which correspond to the images of the patterns illuminated by the different sources of the object, in the different section planes. The processing of these values which depend on the position of each source in a section plane and the distance from this plane, makes it possible to reconstruct the sources contained in each section plane and therefore the image of the object for each plane. cutting.

According to the invention. The transparent elements of each pattern of the mask are identified by a row number and a column number. The mask patterns are rectangular, as are the transparent elements of each pattern. The row and column numbers of each transparent element correspond respectively to two numbers in a succession of pairs of numbers in a two-dimensional series. Pairs of numbers in this series are obtained from successive numbers in a predetermined one-dimensional pseudo-random series. If L denotes the total number of Rows and C The total number of columns, L and C are chosen so that the product LxC is equal to The length of a pseudo-random one-dimensional series, L and C being prime to each other.

The table of FIG. 2 gives in the first column, an example of values of a known one-dimensional pseudo-random series, of q-nary type, tabulated from 0 to 650, used in the method of the invention, to form a two-dimensional pseudo-random series. To construct a two-dimensional pseudo-random series The A (i, j) from a one-dimensional series A (x), we write that the binary content of A (i, j) = A (x) with i≈mod x and j≈mod x_ rs The selected series is said to be q-nary and has the following training rules:

_* n n-1 n-2 q -1 q -1 q ^" '^"> m =; _p =; L = and q = k ^{q_1 Q "* q" 1}

k is a prime number and i is an integer, m represents the number of elements of a mask pattern, p The number of holes or transparent elements, l The correlation of the series by itself. 2

Take for example, m = 651, q = 5 and n = 3.

Advantageously, each pattern is chosen so that the ratio p / m (little different from 1 / q) is less than or equal to ten percent.

The pairs of values of the two-dimensional series corresponding to each ^" value of the one-dimensional series, are given in the following two columns i and j of the table in FIG. 2. This is how the value 1 of the one-dimensional series, correspond the pairs of values i = 1 and j = 1 of the two-dimensional series. A The value 47 correspond the values i = 16 and j = 5 etc. In fact, each value i in the two-dimensional series corresponds to the division oduLo 31 of the corresponding value in the one-dimensional series. Each value J of the two-dimensional series corresponds to the division moduLo 21 of the corresponding value of the one-dimensional series.

The values of i and J represent the row and column numbers at the intersection of which, according to the invention, the transparent elements of a pattern, of rectangular or square shape, are placed. The rest of the design is opaque.

A pattern used in the implementation of the method of the invention is represented in FIG. 3. The transparent elements are represented by square or rectangular surfaces, of black color. The i and J numbers of rows and columns are shown in the figure.

To constitute the pattern of FIG. 3, only the values i comprised between 0 and 20 and J comprised between 0 and 30 have been used in the table in FIG. 2. The size of the mask is chosen for example, so that the largest projection of a pattern, fits into the diameter of the detector. Simple calculations show that the signal-to-noise ratio S / B is of the form S / B = kC ι! / F ^" rô 1.

In this expression C is the total value of counts of the detector, k is the ratio between the activity of a source in the object to the total activity of the sources of the object (value of the count for a predetermined time), m represents the total number of opaque elements in a mask pattern, and p is the number of transparent elements (which are actually windows or holes). This relationship shows that at constant total count, the S / N ratio increases when the number of holes decreases. The mask of the invention in which the number of transparent elements is small, provides an S / N ratio which is y 2 times higher than Masks whose transparency is close to 50%, for example of the type of those used for FENIMORE and CANNON and does not require a homogeneous response from the camera, better than the percent.

We will show more precisely, the influence of the transparency of the mask, and in particular of the low transparency, in the improvement of the signal / noise ratio. If the pattern comprises, as indicated above, m elements among which there are ap holes or transparent elements, to obtain a signal at the output of the detector, it is necessary to add p pixels and to subtract therefrom (mp), affected by the coefficient l / (pL ) (l denoting The correlation of the series by itself). The contents A of pixels (number of strokes for these pixels) differ little from each other.

We can write, for the variance of the signal:

The ratio if gna l / noise can be written: p. O S / B = - £ =, in Which 0 is The number of blows received from an object element and crossing a hole in the mask. It is possible to establish a relationship between A and 0, according to BUDINGER (Positron Emission

To ography - Alan P. LISS INC. 1985 - page 234). We. admits that 20% of the object volume contains radioactivity.

We take for 0 a value corresponding to the mean of the active object elements and we admit that t = 12 slices are discernible. We have :

A = 0.2pxtx0 = 2.4 pO

In the case of masks with 50% transparency

m-p

2 p (p-l) is little different from 1. This expression is very small in front of 1 for a pattern of 651 elements. Hence the improvement of 2 in the S / N ratio when the transparency of the mask decreases.

Replace 0 with A and put:

. L ² m — p

1 + x = in relation (1)

(pl) ²

We obtain :

The number of strokes per pixel increases as ^ (p. Let us take as an example: a pattern having

.and a pattern having m = 31x21 elements, with p = 26 holes and ^ = 1. The relationship between the respective values of A is

39. But due to fluctuations in pixel size the

2 2 noise fluctuations must be multiplied by (1 + A (C "/ S ⁾⁾ _> where <5" 7S is the relative fluctuation on the size of the pixel. For -2 2 2 4 r7S = 0.5.10, AÇσ ^~~ / S) = 1 for A = 4.10, which corresponds to a signal / noise ratio of 1.8 with a mask of 50% transparency and of 12 with a mask of 4% transparency. We therefore see that, due to practical considerations, the second type of mask is much better suited to specifying the content of the object.

To reconstruct the images of different sections of the object, the procedure for the invention is as follows, for one of these images: the count values supplied by the camera are recorded in a memory; these values of counts correspond to the radiation received by the detector, through the transparent elements of one or other of the patterns of the mask, when the latter is illuminated by the different sources of the different cutting planes. For the processing of these recorded values, the procedure is as follows. The count values are selected according to the decoding matrices deduced from the coding matrices of the positions of the transparent elements of a pattern. These decoding matrices correspond to the images on the detector, of at least one pattern of the mask illuminated by the sources contained in the object. As indicated above, the dimensions and locations of these images on the detector depend on the coordinates of the sources in each section plane, as well as on the distance of each plane relative to the detector. In one embodiment of the method of the invention, one chooses for example decoding matrices corresponding to the images of the patterns which are centered on the axis of the detector, these images occupying a portion of the detector, - in the center of that -this. It follows that to obtain images occupying this central position, it is necessary to consider that one or the other of the patterns of the mask, are illuminated by the sources, the decoding matrices being chosen accordingly.

The process then consists in carrying out the following operations: - as indicated above. top, we select a portion of the detector for each distance from the cutting plane Pi with respect to the detector, - for each cutting plane Pi, we move the decoding matrix to identify the sources possibly present in this plane Pi, this displacement corresponding to a scan of all the sources identified by their coordinates (i, j) in the plane Pi, and by measured digital values which are recorded in memory, the sources identified being, according to this embodiment, those located on straight lines passing through the geometric center of the detector. - from the corresponding numerical values, the activities of the sources of each plane are reconstructed. The values of these activities are saved in memory.

According to this embodiment of the invention, the following operations are then carried out:

- For each plane Pi away from predetermined ent, a comparison is made between the activity of each source of coordinates (i, j) thus reconstructed in the plane Pi, and the respective reconstructed activities for neighboring sources located in neighboring planes Pi + 1 and Pi-1, having the same coordinates (i, j) as the source considered in the plane Pi,

- it is determined whether the differences between the activity of each source of coordinates (i, j) in each plane Pi and the respective activities of the sources of the same coordinates (i, j) in the neighboring planes Pi-1 and Pi + 1 exceed a predetermined threshold, this exceeding indicating the location of a source of coordinates (i,) in the plane Pi,

- an activity is assigned to each source thus identified; this activity and the coordinates of the source are saved in memory,

- the count value that this source would produce from the activity assigned to it is calculated for each identified source. This calculated value is stored in memory, - for each source identified, we subtract the count value actually measured previously during scanning ⁽ for the source with the same coordinates Ci, j) in the same plane ⁾ , from the value of counts calculated for the identified source, so as to obtain a corrected count value for the identified source. This corrected value is saved in memory,

- finally, the activity of each identified source is determined from the corresponding corrected counting value and the preceding treatments are repeated in an iterative mode until the corrected value obtained for each source be invariable. It is from the invariable corrected values that the images of the sections of the object are obtained. The iterative processing in fact stops when the signal / noise ratio reaches a sufficient value, that is to say when the background noise has become insignificant.

FIG. 4 schematically represents a system for reconstructing images of successive parallel sections of an object or an organ containing sources emitting gamma radiation and making it possible to implement the method of the invention.

This system includes a gamma-camera 5, located opposite the object or the member 2 for which one wishes to reconstruct images of successive parallel sections. This gamma-camera comprises in particular a detector 1 (without collimator) which is not shown in detail in the figure, but which comprises in known manner, a scinti llator associated with photomultipliers. These photomultipliers are connected to supply, amplification and coding means 6, which supply, on a plurality of outputs 7, a plurality of coded values of the energy of the radiations received by the detector. These values are coded so that each value corresponds to information relating to the coordinates of the point of impact of each radiation on the detector. These coordinates are identified in the reference frame mentioned above. Also shown in this figure is the mask M comprising an assembly of patterns such as M1, described above. This mask is interposed between the detector 1 and the object or the organ 2. In this system, the gamma-camera does not include a collimator but is provided with a field of view limiter 8. The coded values supplied by the outputs 7 of the camera are recorded in a memory 9. The outputs of this memory are connected to decoding means 10 of the stored coded values. These decoding means are connected to a processing unit 11 which applies to them the decoding matrices mentioned above. The processing means receive the values thus selected according to these matrices. The decoded values can be recorded in a storage unit 12, connected to the processing means 11, before being processed so as to then be transmitted to display means 13 of the reconstructed images of the different cutting planes of the object or of the object. 'organ 2. The storage means 12 also make it possible to record the programs defining the decoding matrices of the coded values and stored in the memory 9. The storage means 12 can also record the programs for processing the decoded values; these programs are not described here in detail. They implement a flowchart in accordance with the steps of the method described above.

The device gives a signal to noise ratio and a resolution all the better as the source is more surface. An appreciable improvement in the detection of deep sources consists, according to another embodiment, in taking on the detector at least two projections of a pattern of the mask as far apart as possible. Two advantages are obtained: one due to the obliquity of gamma rays participating in Image reconstruction, the other linked to counting statistics by increasing the detection area used. The advantages of an oblique projection involving at least two projections of a pattern on the detector are as follows: 1) - the signal / noise ratio for a given mask and mask-detector distance, increases in proportion to the square root of the detection surface. If we consider two unrelated projections, the gain in S / B is V ~, it is γ2 / (1-K * 0 if the two projections have a relative common part ^ ^{For a} magnification (ratio between the dimension of a pixel and that of a hole) varying from V2 of the superficial slice, to the deepest slice. The S / N ratio remains constant.

2) - The skew effect of gamma rays improves localization in. depth. This location depends on the degree of overlap between the projections of the holes seen from the object element. This overlap is all the weaker as the projections are more oblique. The phenomenon is comparable to that of Telemetry. The larger the base, the better the appreciation of the distance. The largest basis is to occupy the entire surface of the detector. It is easy to show that the depth location for a given diameter of the detector increases with the number of pixels in a diameter. Depending on the desired lateral and depth resolutions, we therefore have to choose two or more projections on the detector.

Claims

CLAIM J5. 1. Method for reconstructing images of successive parallel sections of an object (2) containing sources emitting gamma radiation, using a gamma camera ⁽ 5 ⁾ with plane detector (1) parallel to the planes of sections, this camera having a plurality of outputs (7) each identified by a code and corresponding respectively to the different possible impact coordinates and to the energy of the gamma rays received by the detector, this energy being measured by a count value supplied by the output corresponding to an impact, the method consisting in interposing between the object and the detector, parallel to this detector, a plane mask (M) formed by assembling identical patterns (M1), of rectangular shape and each comprising a mosaic transparent elements in a surface opaque to radiation, these transparent elements occupying in the pattern predetermined positions identified according to a coding matrix, the re construction of the images of the sections being carried out by processing the count values recorded in a memory (9), these values being selected according to decoding matrices deduced from the coding matrices and corresponding to the images on the detector (1 ⁾ of at least one pattern (M1) of the mask illuminated by said sources. The dimensions and locations of these images on the detector (1) depending on the pattern chosen in the mask, the coordinates of the sources in each plane, with respect to a reference frame, as well as the distance of each plane from the detector, the process consisting in carrying out the following operations:

a portion of the detector is selected for each distance from the cutting plane Pi relative to the detector,

- for each cutting plane Pi, the decoding matrix is moved to locate the sources possibly present in this plane, this displacement corresponding to a scanning of all the sources identified by their coordinates (i, j) in the plane Pi, and by measured count values, the sources identified being those located on lines passing through the geometric center of the detector,

- from the corresponding count values, the activities of the sources of each plane are reconstructed, characterized in that the following treatments are then carried out:

- it is determined whether the differences between the activity of each source of coordinates (i, j) in each plane Pi and the respective activities of the sources of the same coordinates (i, j) in the neighboring planes Pi-1 and Pi + 1 exceed a predetermined threshold, this exceeding indicating the location of a source of coordinates (i, j) in the plane Pi, - an activity is assigned to each source identified,

- the count value that this source would produce from the activity assigned to it is calculated for each identified source,

- we subtract, for each identified source, the count value actually measured previously for the source with the same coordinates (i, j) in the same plane, from the count value calculated for the identified source, so as to obtain a value of corrected counts for the identified source,

- the activity of each identified source is determined from the corresponding corrected counting value and the previous processing is repeated in an iterative mode until the corrected value obtained for each source is invariable.

2. Method according to claim 1, characterized in that each of said transparent elements of a pattern (M1) is identified by a row number and a column number, the total number of rows L multiplied by The total number of columns C being equal to the length of a pseudo-random one-dimensional series, L and C being prime between them.

3. Method for reconstructing images of successive parallel sections of an object (2) containing sources emitting gamma radiation, using a gamma camera (5) with plane detector (1) parallel to the planes of sections, this camera having a plurality of outputs (7) each identified by a code and corresponding respectively to the different possible impact coordinates and to the energy of the gamma rays received by the detector, this energy being measured by a count value supplied by the output corresponding to an impact, the method consisting in interposing between the object and the detector, parallel to this detector, a mask (M) plane-formed by assembling identical patterns (M1), of rectangular shape and each comprising a mosaic of transparent elements in a surface opaque to radiation, these transparent elements occupying in the pattern predetermined positions identified according to a coding matrix, The rec the images of the sections are constructed by processing the counting values recorded in a memory (9), these values being selected according to the decoding matrices deduced from the coding matrices and corresponding to the images on the detector (1) at least a pattern (M1) of the mask illuminated by said sources, the dimensions and locations of these images on the detector (1) depending on the pattern chosen in the mask, the coordinates of the sources in each plane, with respect to a reference frame, as well as the distance of each plane from the detector, The process consisting in carrying out The following operations:

- for each cutting plane Pi, the decoding matrix is moved to locate the sources possibly present in this plane. this displacement corresponding to a scan of all the sources identified by their coordinates (i, j) in the plane Pi and by measured count values, - from the corresponding count values, the activities of the sources of each plane are reconstructed, characterized in that each of said transparent elements of a pattern (M1) is identified by a line number and a column number, the total number of lines L multiplied by the total number of columns C being equal to the Length of a pseudo-random one-dimensional series, L and C being prime to each other, the two-dimensional series The A (i, j) being obtained from the one-dimensional series A (x) by writing that the binary content of x, The selected series

being q-nary and sat its health to the quat ons:

n n-1 n-2 q -1 q -1 q -1 i

—--; p = - ^• l = and q = k q-1 q-1 q-1

4. Method according to claim 3, characterized in that it consists in choosing each pattern so that the p / m ratio (p _eu different from 1 / q) is less than or equal to ten percent.

5. Method for reconstructing images of successive parallel sections of an object (2) containing sources emitting gamma radiation, using a gamma camera (5) with plane detector (1) parallel to the planes of sections, this camera having a plurality of outputs (7) each identified by a code and corresponding respectively to the different possible impact coordinates and to the energy of the gamma rays received by the detector, this energy being measured by a count value supplied by The output corresponding to an impact. The process consisting in interposing between the object and the detector, parallel to this detector, a plane mask (M) formed by assembling identical patterns (M1), of rectangular shape and each comprising a mosaic of transparent elements in a surface opaque to radiation, these transparent elements occupying in the pattern predetermined positions identified according to a coding matrix, the reconstruction of the images of the sections being carried out by processing the count values recorded in a memory (9), these values being selected according to matrices of decoding deduced from the coding matrices and corresponding to the images on the detector (1) of at least one pattern (M1) of the mask illuminated by said sources, the dimensions and locations of these images on the detector (1) depending on the chosen pattern in the mask, coordinates of the sources in each plane, relative to a reference frame, as well as the distance nt of each plane relative to the detector, the process consisting in carrying out the following operations:

- for each cutting plane Pi, the decoding matrix is moved to locate the sources possibly present in this plane, this displacement corresponding to a scanning of all the sources identified by their coordinates (i, j) in the plane Pi, and by measured count values,

6. Image reconstruction method, implementing the steps defined in Claims 1 and 3.

7. Image reconstruction method, implementing the steps defined in claims 1 and 5.

8. Image reconstruction method, implementing the steps defined in claims 3 and 5.