WO2006013585A1 - Method and equipment for monitoring the functionality of the spinal cord - Google Patents

Abstract

A method for monitoring the functionality of the spinal cord of a living creature comprises the steps of: a) acquiring an electric signal (S1) obtained in response to an electric stimulation applied to the creature's body; this signal comprises a first signal (S) indicative of a somatosensory evoked potential generated by said pulse; b) acquiring a second electric signal (S2) indicative of a H-reflex potential generated by said electric stimulation; c) processing the first and second electric signals to simultaneously provide first and second information, which are correlated to the first and second electric signals thus enabling to check the functionality of the spinal cord.

Description

DESCRIPTION

Method and equipment for monitoring the functionality of spinal cord.

The present invention relates to a method and equipment for monitoring the functionality of the spinal cord of a living creature.

The monitoring of the functionality of body organs, and generally the vital parameters of a living creature is known to be a basic aspect in up-to-date surgery techniques. Particularly, in the case of surgery carried out on the vertebral column to correct spinal deformity (for example, serious forms of scoliosis) or to remedy spinal injury, a continuous monitoring of the spinal cord functionality is required. In fact, in the course of a surgery, the integrity and functionality of the cord may be compromised by manoeuvres performed by the surgeon to such an extent as to cause, for example, a prolonged insufficient blood afflux to the cord and, consequently, the temporary malfunctioning of the latter. Sometimes, these manoeuvres can cause a permanent and irreversible damage to the spinal cord and determine, in the most serious cases, the paraplegia of the living creature.

Therefore, monitoring the cord functionality during surgery is aimed at early detecting states of cord distress such as to be able to remedy before any irreversible damage may occur.

The monitoring methods currently used are still at the experimental stage and cannot provide sufficiently accurate results. Particularly, said methods are based on the stimulation of the nervous system with subsequent observation of effects induced by these stimulations which may be related with the functionality of spinal cord. Among the methods used, some are invasive in that they provide the bloody positioning of catheters provided with stimulating electrodes and recorders within the vertebral canal. These monitoring methods are disadvantageous not only because they are difficult to be implemented, but also because they may give rise to cord lesion or infection. Furthermore, sometimes the electrodes cannot be held in position throughout the surgery.

Other methods are not-invasive such as that based on the detection of a somatosensory evoked potential or SEP. According to two further conventional monitoring methods either a motor potential or a H-reflex are detected. These methods are known to those skilled in the neurophysiology field. The monitoring method based on the detection of the somatosensory evoked potential (SEP) is performed by electrically stimulating a peripheral nerve of the living creature's body by means of electrodes placed proximate to the nervous fiber, for example through a transcutaneous stimulation at the popliteal canal (on the rear side of the knee) . In response to this stimulation, the method provides the recording of a electroencephalographic potential (EEG) by means of further electrodes positioned on the scalp proximate to the somatosensitive cerebral cortex.

In greater detail, this monitoring method provides carrying out subsequent series of electric stimulations, each series consisting of at least fifty stimulations, with consequent recording of the cerebral activity at each stimulation in the series, i.e. fifty electroencephalographic potentials (EEG) are recorded corresponding to each stimulation in the series.

Each recorded electroencephalographic potential (EEG) can be assimilated to a background noise which includes the SEP potential, i.e. the signal to be detected. The ' SEP potential corresponding to the series of fifty stimulations can be extracted by averaging the EEG potentials recorded. In fact, the noise, i.e. the EEG potential, superimposed to SEP potential can be eliminated only by assuming it is white. However, the hypothesis of the EEG potential as being a white noise is not a correct one, since it has a not null autocorrelation and further the somatosensitive evoked potential is not a time-invariant process, all the more in long-time surgery.

Moreover, temporary alterations of the cord functionality having a shorter duration than the averaging time cannot be detected by carrying out an averaging operation. The monitoring method based on the detection of the motor potential provides that the motor cerebral cortex is stimulated by means of magnetic pulses. Subsequent to this stimulation, the motor cortex generates nervous pulses, or action potentials, propagating along the nervous pathways to the peripheral muscles. By being excited by the action potentials, these muscles generate muscular electric potentials which can be detected and recorded by means of electromyographs.

The monitoring method based on the H-reflex detection provides a stimulation of a body' s peripheral nerve which controls the sensory and motor functions. In response to this stimulation, the method provides recording the potentials produced by the contraction of the addressed muscular bundles, i.e. the myoelectric potentials. Each of the above monitoring methods is implemented by known, marketed equipment, such as electromyographs, electroencephalographs, electric and magnetic stimulators. Each of the above monitoring methods has drawbacks from the neurophysiologic point of view.

In fact, only the nervous pathways located on the rear side of the cord can be monitored by the SEP evoked potential, while those located on the side and front thereof cannot. Furthermore, the activity of peripheral nervous cells or motoneurons transmitting the motor pulses and located in the cord grey matter cannot be controlled.

A drawback of the method for detecting the motor potential lies in the difficulty of magnetically stimulating inner portions of the motor cerebral cortex, such as the portion controlling the muscles of lower limbs. Accordingly, only the functionality of the cord portion controlling the upper motor pathways of the body can be effectively inspected by this method. For this reason, the motor potential method is applied only for research purposes and is not a clinical routine.

Finally, the monitoring method based on the detection of H-reflex allows to inspect only the ascending and descending neural pathways running within the spinal cord portion located approximately beneath the ninth dorsal vertebra. As a consequence, some functional alterations of the cord portion extending above the ninth vertebra either cannot be detected or can be only indirectly detected.

Furthermore, the results achieved by the above monitoring methods can also change as a function of the living creature being monitored, the anesthesia techniques used, the body temperature and the spinal cord temperature. Accordingly, since the variability of the recorded signal not always depends on alterations of the cord functionality, wrong results (false positives) may be obtained.

The object of the present invention is to provide a method for monitoring the spinal cord functionality allowing to overcome the drawbacks of the above conventional methods.

This object is achieved by a monitoring method such as defined in annexed claim 1. Preferred embodiments of the invention are as defined in the dependent claims 2 to 13.

The object of the present invention is also an equipment implementing said monitoring method such as defined in annexed claim 14. Variant embodiments are as defined in dependent claims 15 to 24. The characteristics and advantages of the present invention will be understood from the following detailed description of an exemplary and non-limiting embodiment thereof, with reference to the annexed figures in which: - Figure 1 is a perspective image of a monitoring equipment according to the present invention;

Figure 2 is a functional block diagram of an inner structure of a portion of the equipment from Figure

1; - Figure 3 is a block diagram of a stimulation module of the equipment from Figure 1;

Figure 4 is a block diagram of an acquisition module of the equipment from Figure 1;

Figure 5 schematically illustrates an exemplary application of the equipment from Figure 1 to a living creature's body to monitor the cord functionality thereof;

Figure 6 illustrates an exemplary neural network implemented by a block of the structure from Figure 2;

Figures 7A and 7B are waveform diagrams of an electroencephalograph^ potential and a somatosensory evoked potential, respectively;

Figure 8 is a waveform diagram of a H-reflex potential. Equipment 100 for monitoring the spinal cord functionality in accordance with the invention is described with reference to Figures 1, 2, 3 and 4. This equipment 100 allows to simultaneously detect and compare electric signals of biological nature or bio-signals excited in a living creature's body (such • as, particularly, a human being) in response to suitable electric stimulations. Particularly, the above bio- signals are a H-reflex muscular potential and a somatosensory evoked potential (SEP) obtained from the recording of an electroencephalograph^ potential (EEG) .

With reference to Figure 1, equipment 100 comprises a main body 8 to manage the monitoring operation and electric connection means 5, 6 and 7 between the main body 8 and the living creature's body (not shown in Figure 1) whose cord functionality is desired to be monitored. These connecting means 5, 6 and 7 are embodied by electric cables having connectors 9, 10, 11, respectively, for connection to the main body 8 on the one end thereof, and on the other end a plurality of metal terminals or electrodes 14-21. Particularly, the metal terminals of cables 5 and 6 are first 14, 15, 16 and second 17, 18, 19 recording electrodes, respectively, whereas the metal terminals of cable 7 are stimulating electrodes 20 and 21. The stimulating electrodes 20 and 21 are electrically isolated from the recording ones in order to supply electric stimulations to the living creature to be monitored. The first recording electrodes 14-16 comprise an anode 14 and a cathode 16 to detect a first electric signal Sl referred to a common electrode 15. This first signal Sl is an electric analogue signal indicating an electroencephalograph^ potential EEG from which the evoked potential SEP can be extracted according to modalities which will be described below.

The second recording electrodes 17-19 comprise the anode 17 and cathode 19 to detect a second electric signal S2 referred to the common electrode 18. This second signal S2 is an electric analogue signal indicating the H-reflex muscular potential.

Moreover, the main body 8 comprises further connectors 12 through which equipment 100 can be connected to a supply voltage and advantageously to further external equipment (such as mass-storaging devices) or to control interfaces (e.g., mouse, electronic processor keyboards, etc.) .

It should be noted that in a preferred embodiment, equipment 100 comprises a touch screen 13 to promote a quick communication between equipment 100 and an external operator (e.g. sending commands to the equipment or receiving information therefrom) who will simply have to touch specific areas on the screen 13 to send selections, command signals, etc. to the equipment 100. Moreover, the monitoring results can be viewed on screen 13.

Preferably, an inner structure 200 of main body 8 of equipment 100 can be provided in a plurality of functional blocks or modules, either of the analogue or digital types, such as schematically illustrated in Figure 2. In other words, some of the above blocks are provided as circuital modules, i.e. with a hardware, whereas others are provided as software modules, i.e. program instruction sequences.

Structure 200 comprises a main block or system control unit 22 (UCS) . This unit 22 is, for example, a microprocessor, a microcontroller or a DSP (digital signal processor) intended to run a dedicated program, i.e. a series of machine instructions implemented therein. The system control unit 22 interacts with a stimulation module 24 (MOD STIM) by exchanging with it first information 23 of a numerical type to control the delivery of the electrical stimulus to the living creature's body to be monitored. Particularly, the exchanged information 23 are bit flows relating to characteristics of the electric stimulus such as width, frequency and duration of the stimulus as well as data indicating the amount of electric charge supplied per each stimulus. Moreover, structure 200 comprises a first 27 (MOD ACQl) and a second 28 (MOD ACQ2) acquisition modules of first Sl and second S2 electric signals, respectively. Particularly, these acquisition modules 27 and 28 are electronic devices comprising circuits (not shown in detail in Figure 2) to receive and filter the above signals and in which the parameters (amplification gains, cut-off frequency) can be modified by unit 22 through control instructions 25 and 26 of a numeric type (bit flows) . Furthermore, a first SdI and a second Sd2 digital signals on output from the acquisition modules 27 and 28 are obtained by performing an analogue/digital conversion of the electric signals Sl and S2. Particularly, the digital signals SdI and Sd2 are bit sequences codifying samples of the first Sl and second S2 electric signals, respectively.

These signals SdI and Sd2 are the input information of the control unit 22 and, particularly, of a first 33 (MOD VAL 1) and a second 34 (MOD VAL 2) evaluation module of the signal quality, respectively. Preferably, the evaluation modules 33 and 34 are made in logic blocks of a program implemented by the system control unit 22. Particularly, unit 22 interacts with modules 33 and 34 through first 31 and further first 32 instructions to modify procedures and evaluation functions, respectively, and through second 31' and further second 32' instructions to receive the evaluation results. The first 31, 32 and second 31', 32' instructions are information of the digital type.

It should be observed that first 33 and second 34 evaluation modules can be also provided within a hardware structure, i.e. by means of an electronic device, such as a DSP.

The first 33 and second 34 evaluation modules can be connected to first 35 (ELAB DIG 1) and second 36 (ELAB DIG 2) digital processing modules of digital signals SdI and Sd2, respectively, by means of logical switches managed by activating commands 201 of the system control unit 22. These processing modules 35 are 36 logic blocks of a program implemented by the control unit 22 to process the acquired signals. The results of the above processing are provided at the control unit 22 by means of further instructions 37 and 38 of the digital type.

Furthermore, a memory module 40 (MEM) is connected to the system control unit 22 to store first numerical data 39, i.e. bit flows corresponding to results of the processing of signals SdI and Sd2. For example, these storing means are embodied in a non-volatile memory of the flash type.

Finally, a display module 213 (VIS) is connected to unit 22 to receive second numerical data 41 to be displayed. Such as illustrated with reference to Figure 1, the display module 213 includes the touch screen 13 helping the quick transmission of external commands 42 to the control unit 22. A preferred embodiment of the stimulation module 24 of the invention is described with reference to the block diagram of Figure 3. It should be observed, particularly, that the stimulation module 24 is an electronic device comprising digital and analogue circuits each of which is represented by a block within said diagram.

The module 24 comprises a digital control unit 41 (CONT DIG) to control and interpret the first numerical information 23 coming from the system control unit 22. The digital unit 41 interfaces with a stimulation isolation unit 43 (UIS) of the analogue type through a digital-analogue converter 42 (D/A CONV) . Through said converter 42, a reference analogue voltage 240 is transferred to the isolation unit 43 to control the amplitude and waveform of the generated electric stimulus. A timing digital signal 241, directly supplied by the digital unit 41 to the isolation unit 43, allows to adjust the frequency and duration of the electric stimulus.

The configuration of the output stage of the isolation unit 43 may change depending on the stimulation mode employed. In fact, it is possible to perform either current stimulation or voltage stimulation.

In the case of current stimulation, the isolation unit comprises a high-impedance output stage outlined by block 44. For example, this output stage is embodied by a DC-DC isolation converter to transfer a low supply voltage to a high voltage. Moreover, in block 44 there is provided at least a power transistor which can be activated by the transferred high voltage in order to act as a current generator and supply the desired stimulus current.

In the current stimulation configuration, the stimulation module 24 can further comprise a suitable circuitry (not illustrated in Figure 3) to calculate the amount of electric charge supplied with the stimulus and send the result of this operation to the system control unit 22 in a digitalized form.

In the case of voltage stimulation, the output stage of the isolation unit 43 is a low-output-impedance stage comprising for example an isolated power transformer outlined by block 45.

In the voltage stimulation configuration, module 24 comprises a current measuring circuit 46 (MIS CORR) to detect the current delivered following the voltage stimulation. This measuring circuit 46 is for example a amperometric clamp comprising an isolation transformer.

Moreover, the stimulation module 24 comprises an integration circuit block or integrator 47 (INT) and an analogue/digital converter 48 (A/D CONV) to evaluate the amount of charge corresponding to the measured current and supply such information to the digital control unit 41 in a digital form, respectively. Thereby, the data referring to the electric charge supplied with the stimulus are provided to the system control unit 22. Referring to Figure 4 there is described a circuit architecture 400 of the acquisition modules 27 and 28. Particularly, this architecture 400 is in the form of a block diagram.

First of all, the circuit architecture 400 of each acquisition module 27, 28 comprises electrodes 49 and 50 to record signals relative to a common electrode 51. This common electrode 51 is indicated by a ground potential and placed away from the recording spot.

Particularly, the electrodes 49-51 correspond to electrodes 14-16 from Figure 1 if the architecture 400 implements the first acquisition module 27, whereas if it implements the second module 28 they will correspond to electrodes 17-19.

Moreover, electrodes 49 and 50 are connected to a high-impedance differential input stage 52 (ST IN) . For example, this input stage 52 is provided by a low-noise operational amplifier in a buffer configuration.

The input stage 52 is connected to a first analogue filter 53 (FIL AN) of the high-pass type having a modifiable cut-off frequency. For example, in the case of first module 27 the cut-off frequency of the filter is 0,01 Hz, whereas in the case of second module 28 it is 10 Hz.

In addition, the architecture 400 provides a differential amplifier 54 (AMP) cascade-connected to the first filter 53 and having a high common mode rejection ratio CMRR. A preferred value of the common mode rejection ratio is 120 dB. Moreover, the amplification factor of amplifier 54 is different for the first 27 and second 28 acquisition module. In the case of first module 27 the amplification factor is, for example, 10.000 (80 dB) , whereas in the second module 28 the amplification factor is, for example, 100 (40 dB) .

After the amplifier 54 a second analogue filter 55 (FIL) of the low-pass type is provided. For this filter 55 the cut-off frequency can be also modified, and may- take values for example of 0,1 kHz and 10 kHz for first 27 and second 28 modules, respectively.

It should be noted that for both the first 27 and second 28 modules, the values of the cut-off frequencies of filters 53 and 55 as well as the amplification factor of amplifier 54 are managed by the system control unit

22.

Advantageously, all blocks described in the circuit architecture 400 have a ground reference potential in common and are connected to the common electrode 51.

These blocks are a "part applied to the living creature to be monitored" .

Advantageously, an electric isolation is provided, which is outlined in Figure 4 with an isolation block 56

(IS), to uncouple the "applied part" from the remaining energized parts of equipment 100. Particularly, this isolation block 56 is situated on output of the applied part to supply a isolated supply voltage 58 to the circuit blocks of the applied part.

Finally, the isolation block 56 is connected to an input of a further analogue-digital converter 57 (A/D) having, for example, a 12 bit resolution.

As it is well known to a neurophysiology expert, the generation of the electroencephalograph^ potential EEG and H-reflex in a living creature's body by means of electric stimulations involves main nervous pathways which are anatomically and functionally independent from one another. Therefore, in general, the bio-signals of H- reflex and evoked potential SEP which can be deduced by the potential EEG, change over time in an independent manner to one another.

However, it has been observed that the simultaneous generation of both bio-signals also involves the same periferal nerve endings and that each of them depends on the integrity of the spinal cord nervous pathway. Moreover, it has been perceived that a state of cord distress, such as a cord lesion, influences both of these bio-signals (SEP and H-reflex) and that, advantageously, this state can be detected by noticing a simultaneous variation of both bio-signals.

A monitoring method of the spinal cord functionality of the invention provides, first of all, the simultaneous detection of first Sl and second S2 electric signals indicating the electroencephalograph^ potential EEG and the H-reflex muscular potential, respectively. Particularly, said bio-signals EEG and H-reflex are generated in the body of the living creature to be monitored in response to the same electrical stimulation. For example, this electrical stimulation may be one single electric pulse, a train of electric pulses, or biphasic electric pulses.

For clarity reasons, in the following, this electrical stimulation will be assumed as being one single electric pulse.

Advantageously, equipment 100 allows to perform monitoring in accordance with the above method. In fact, this equipment 100 provides suitable electric pulses to stimulate the living creature's body to be monitored by detecting and processing the electric signals Sl and S2 thus generated in real time.

An exemplary functioning of the inventive monitoring equipment 100 is described with reference to Figures 5, 6, 7A, 7B and 8. First of all, equipment 100 is connected to the living creature's body to be monitored by means of the first 14-16 and second 17-19 recording electrodes and by means of the stimulation electrodes 20 and 21.

Referring to Figure 5, the living creature to be monitored has been schematically represented by encephalon 501, a vertebral column 502 (i.e. the spinal cord) and a leg 503.

The stimulation electric pulse is supplied transcutaneously, by positioning the stimulation electrodes 20 and 21 of equipment 100 on a knee 504 of leg 503. Particularly, one of both electrodes (for example, electrode 20) is positioned on the posterior side of knee 504 proximate to a popliteus muscle and a posterior tibial nerve, whereas the other one (now shown in Figure 5) is placed on the front side of knee 504.

This pulse supplies a suitable amount of electric charge thus selectively stimulating both nervous fibers connected to a soleus gastrocnemius muscle group of leg 503 and cutaneous sensitive fibers of the posterior side of knee 504.

An action potential, i.e. a nervous pulse generated by the electric stimulation propagates through the body towards the central nervous system via the above- mentioned nervous fibers. This propagating step is outlined by a first path 506 of Figure 5.

Subsequently, at the fifth lumbar vertebra or the first sacral vertebra, the stimulated nervous fibers enter the vertebral canal (not shown in Figure 5) of vertebral column 502 such that the action potential being generated may propagate in the direction of the encephalon 501 thus exciting the sensitive fibers arranged on the right or left side (according to the stimulated leg) of cord.

At the ninth-tenth dorsal vertebra, these sensitive fibers when stimulated excite a population of peripheral nervous cells of their own or motoneurons present in the cord grey matter located in the front part thereof. Particularly, the motoneurons give rise to a reflex response to the electric stimulus supplied with electrodes 20 and 21. This reflex response is a further electric potential the characteristics of which depend on the parameters of the electric stimulation and the functionality of the nervous pathways involved.

This further potential produced by the motoneurons propagates along nervous pathways within the body, such as schematically shown by a second path 507, until stimulating the soleus-gastrocnemius muscle of leg 503. The excitation of the soleus muscle following the stimulation generates the H-reflexed bio-signal which can be detected in the form of electric potential. Particularly, this electric potential is the second signal S2 detected by means of the second recording electrodes 17-19 positioned at a first recording site 517 proximate to a calf 509 of leg 503. Moreover, it should be noted that the nervous electric potential generated by the electric pulse applied to the knee 504 is transmitted to the spinal cord through cutaneous sensitive fibers and by further nervous fibers which are not involved in the generation of the H- reflexed bio-signal. At the first or second sacral vertebra, said fibres enter the vertebral canal and move forward to the encephalon 501 such as schematically shown by a third path 510. At the encephalon 501, the action potentials being transmitted stimulate a portion of the sensitive cerebral cortex. In response to this stimulation, the cerebral cortex generates the somatosensitive evoked potential SEP which sums up to the electroencephalograph^ potential EEG. This sum of potentials is the first electric signal Sl which can be detected by means of the first recording electrodes 14-16 positioned at a second recording site 514 on the living creature's scalp.

Particularly, it should be observed that EEG potential can be assimilated to a background noise superimposed to the evoked potential SEP which one desires to analyse. As a consequence, the first electric signal Sl should be suitably processed to be indicative of the evoked potential SEP. On the contrary, second signal S2 should not be subjected to excessive processing to supply the information required on the H-reflexed potential.

The processing modes of said signals will be discussed below. Once they have been detected, the electric signals Sl and S2 are amplified and filtered at the first 27 and second 28 acquisition module, respectively (Figure 4) . Particularly, the amplification factor of the differential amplifier 54 used in modules 27 and 28 is different for the first Sl and second S2 signal in that these signals have different amplitudes. In fact, the amplitude of the first electric signal Sl is in the range of μV (for example, 30-40 μV for potential EEG) , whereas the amplitude of second signal S2 is in the range of mV (for example, 1-10 mV for H-reflexed signal) . On output from amplifier 54, the peak amplitude of each signal is thus optimized to favour the proper functioning of the analogue/digital converter 57.

On the other hand, the components of signals Sl and S2 of lower or higher frequencies than those of interest are eliminated by the analogue filters 53 and 55, respectively. In fact, these frequency components may be originated by external disturbances, in which case they are not significant in the subsequent processing step. Subsequent to amplification and filtering, the analogue electric signals Sl and S2 are sampled and quantized by converter 57. Quantization levels of the samples per each signal are coded by means of bit sequences (for example 12 bit) to be sent to first 33 and second 34 evaluation modules as the first SdI and second Sd2 digital signals, respectively. The samples of these signals are contained in a time interval or recording "window" having a preset, though modifiable duration. This time interval is known to those skilled in the art as the recording sweep.

For example, in the case of the first signal Sl indicating the EEG potential (i.e. the SEP potential) the recording sweep has a duration of 125 ms and contains 320 signal samples. In the case of second signal S2 indicating the H-reflex the sweep has, for example, a duration of 62,5 ms and contains 640 signal samples.

In addition, by means of the modules 27 and 28, equipment 100 allows to acquire a further first Sl' and a further second S2' electric signals which are test signals indicating the encephalographic potential bio- signal EEG and an electromyographic muscular potential bio-signal EMG, respectively. Contrarily to signals Sl and S2 detected following the electric stimulation, these test signals Sl' and S2' indicate the potentials EEG and EMG existing in the living creature's body being monitored regardless of the stimulation pulse being sent. To the purpose, a suitable timer of equipment 100 sets the time instants for the acquisition of said signals. Preferably, signals Sl' and S2' are acquired every 10-15 seconds. The recording sweeps containing the signal samples Sl' and S2' are analysed by the evaluation modules 33 and 34 (Figure 2) to check that the potential EEG and EMG are not temporarily altered or degraded by external disturbing signals (such as caused by employing a electrosurgery unit during a surgery to the vertebral column) . In fact, the processing of altered or modified signals may supply wrong and contradictory information on the cord functionality thus nullifying the objective of an early diagnosis of states of cord distress.

Referring to the analysis of the samples of further first Sl' and second S2' electric signals, the first 33 and second 34 evaluation modules check that these signals meet preset principles or evaluation parameters. Particularly, module 33 checks that the amplitude of said signal Sl' does not exceed maximum and minimum values, i.e. this amplitude always takes values comprised within a preset variation range. For example, this variation range is -I V - I V. Furthermore, the amplitude of the low-frequency oscillation of signal Sl' , takes place if, at the signal baseline, excessive ripples are present. Subsequently, the presence of too abrupt oscillations of the acquired signal is examined by means of the analysis of the first derivative of the same signal . Finally, the amplitude is checked of those high- frequency components of signal Sl' which have not been eliminated by the analogue low-pass filter 55.

If signal Sl' meets all above evaluation principles, module 33 sends a stimulation-starting signal to the system control unit 22 through second instructions 31' .

Thereby, the EEG potential is acquired to be processed to extract the evoked potential SEP contained therein.

On the contrary, if signal Sl' does not meet even only one of the above evaluation principles , this means that it is temporarily corrupted by an external disturbing signal and sending an electric stimulus to take the EEG potential is not opportune. In this case, second instructions 31' sent to unit 22 contain a wait signal for the temporary inhibition of stimulation. Similarly, referring to second signal S2', module 34 checks that the amplitude of said signal always takes values comprised between a preset variation range (for example, -IV - IV), i.e. such as to avoid the saturation of the acquisition stage. If signal S2' meets this evaluation parameter, module 34 sends a stimulation starting signal to the system control unit 22 through further second instructions 32' . In this case, a H-reflex potential can be acquired. On the contrary, these further second instructions 32' contain a wait signal.

Following the receipt of the stimulation starting signals by both evaluation modules 33 and 34, the control unit 22 commands the stimulation module 24 through first information 23 for sending a stimulation pulse.

As the system control unit 22 is able to process digital data at high frequency (such as in the range of hundreds MHz) , the evaluation of the quality of a single sweep relative to acquired signals Sl' and S2' is a very quick operation. In this case, it is legitimate to assume that the recording sweeps relative to the electric signals detected following this pulse (such as the above Sl and S2 signals) are exempt from disturbances and hence processable. Accordingly, the control unit 22 transfers these sweeps to the digital processing modules 35 and 36 through the activation command 201.

Moreover, it should be noted that each stimulation pulse can only be applied to the living creature's body after a suitable time has passed from the previous pulse. In fact, as it is known to those skilled in the art, after each stimulation, the nerve centers controlling the generation of the H-reflex bio-signal require a recovery time of about 6-7 seconds to return to optimum conditions to generate a new bio-signal. Advantageously, on operation of the inventive equipment 100, a time interval of about 10 seconds between one stimulation delivery and the next one is provided.

With reference to the operation of the first processing module 35, this module 35 processes the digitalized samples (i.e. the first digital signal SdI) of first signal Sl indicating the electroencephalograph^ potential EEG in order to extract the evoked potential SEP bio-signal to be analysed and displayed.

In this processing, it is assumed that the detected electric signal Sl is the sum of the evoked potential SEP

(designated below with S) and the noise represented by the electroencephalographic potential EEG (designated below with Ns) . Furthermore, it is assumed that the evoked potential S and noise Ns are uncorrelated. Generally, the processing module 35 reconstructs the evoked potential S relative to a single recording sweep starting from a sum of mathematical functions, for example, of the gaussian type. Preferably, to carry out this processing, the module 35 implements a multi-layer Radial-Basis Function Neural Network 600 such as schematically illustrated in Figure 6. Particularly, the neural network 600 is a feedforward unbiased neural network the general characteristics of which are known to those skilled in the art. This network 600 comprises three layers. An input layer 601 includes an input vector X the elements of which are the discrete time instants of each sample of the digital signal SdI. Hence, if the recording sweep contains M samples, this input vector is X=[I 2 ... k ... M] where 1 is the instant of the stimulation pulse and M is the final instant of the single sweep.

The vector X is sent to N neurons 604 of a hidden layer 602. For example, the number of neurons in the hidden layer 602 is N=25. To each neuron 604 of the hidden layer there is associated an activation function having radial symmetry, for example, a gaussian function designated with N(μ-j,σ) (with j varying between 1 and N) . These gaussian functions have their respective centers, being coincindent with the mean values μ_j, located within the time interval of the single recording sweep. Furthermore, said gaussian functions have different mean value μ_j from one another, i.e. the gaussian centers are located in fixed points different from one another within the single recording sweep. For example, 25 gaussian functions can be set per single 125 ms recording sweep, i.e. one every five ms. All gaussians of neurons 604 have the same spread σ.

Moreover, the neural network 600 comprises an output layer 603 to perform a sum of neuron outputs of the hidden layer 602 weighted through weights Wj (with j varying from 1 to N) . The output of the neural network can be calculated as:

Λ-μ _j

Y⁽ⁱ⁾(X) =∑w⁽ _J ¹⁾e ° (1)

where the index i denotes the i^th sweep taken at the i^th stimulus. In (1) with Y(X) is indicated the estimated value of the evoked potential S, whereas μ_j and σ² are the mean and variance of a gaussian-type function associated to a j^th neuron in the network. Furthermore, w^(l)j is the weight attributed to the output of the same neuron for the i^th recording sweep.

The (1) can be expressed in a matrix form, that is: _γ(i) _{= H}τ_w(i) ₍₂₎

where H is a N X M-sized matrix in which each line is a time sequence of the M outputs of the j^th neuron of the hidden layer 602,

HMH₁... Hj ... H_N] (3)

H_j = [H_j(l)...H_j(k)...H_j(M)] (4) whereas W^(l> is a column vector containing the N neural weights of the i^th experimental test.

The overall operation of the network depends on the neuron number N of the hidden layer, on the location of the gaussian centers μj and on the spread value σ.

For example, the neuron number N of the hidden layer is correlated to the shape of the signal to be detected, particular reference is made to the number of signal peaks.

With reference to the distribution of centers μ_j of gaussian functions, it has been choosen to set these centers considering the following relation:

where j varies from 1 to N. According to (5) the gaussian centers are a function of the number N of gaussians and the extension M of the time sweep.

As it is known to those skilled in the art, a basic parameter related to the monitoring of the spinal cord functionality is the time interval of a peak occurrence in the waveform of the evoked potential S which is defined as signal latency. It has been observed that the fixed distribution of the centers of the gaussian functions in accordance with (2) is misleading for the evaluation of the latency. For this reason, it has been chosen to adaptatively change the center μ_j* of a gaussian while maintaining the other centers fixed.

On the contrary, the common value of the gaussian spread σ (i.e. of variance σ²) is preset and is not changed over processing since this does not degrade the ability of the neural network 600 to approximate the evoked potential S. The spread σ is linked both to the neuron number N of the hidden layer and the sample number M of the sweep according to the relation:

o=β(^] (6)

where an experimental parameter β is set relative to a compromise between the ability to follow changings both at high-frequency and low-frequency of the signal to be detected. Particularly, the amplitude of the gaussian used will be greater or lower based on a greater or lower value than β. Moreover, the parameter β is fixed for all gaussians unless an external action is performed to change it.

The detection of the evoked potential S is performed by instructing the neural network sweep by sweep, i.e. adaptatively updating the weights W^(l) on output from each neuron and the location of the center μ-j_* of a gaussian function.

Advantageously, this is achieved by the minimization of the mean square error between the estimated output of the neural network Y and the detected signal SdI. Particularly, at the i^th stimulation, the output of the neural network Y^(l) and the detected signal SdI^(l) (corresponding to the sum of the potential S^(l) and the electroencephalograph^ noise Ns^(l) both of them digitalized) are compared and the mean square error is calculated. Particularly, this error is expressed in a matrix form as:

E [E^(i)TE⁽ⁱ⁾] =E [ (SdI⁽ⁱ⁾-Y⁽ⁱ⁾ )^τ(SdI⁽ⁱ⁾-Y⁽ⁱ⁾ ) ] = E[(S⁽ⁱ⁾+Ns⁽ⁱ⁾-Y⁽ⁱ⁾)^T(S⁽ⁱ⁾+Ns⁽ⁱ⁾-y⁽ⁱ⁾)] (7)

Subsequently, the adaptation is performed in two steps: the first one is connected to the updating of weights W^(l) and the second one is connected to the displacement of center μ-j_*. With reference to the adaptation of weights, since S^(l) and Ns^(l) are uncorrelated, the expression of the mean square error (7) can be modified as follows:

E[E^(i)TE^<i>]=E[(S⁽ⁱ⁾-Y⁽ⁱ⁾)^T(S⁽ⁱ⁾-Y⁽ⁱ⁾)]+E[Ns^(i)TNs^(1>] (8) As E [Ns^(l)TNs^(l>] can be considered as being independent from W^(l) and the placement of the centers, the mean square error is minimized when E[Y^(l)]-^ S^(l>.

The weight convergence is performed by using a Least Mean Squares algorithm or LMS expressed in a matrix form:

_W ⁽ⁱ ₊ ^D _{= w}<ⁱ>₊₂ηH(Sdl^(i>-Y⁽ⁱ⁾) (9) where η e is a convergence rate. The (9) indicates that the updated weight w^(l+1> depends on the weigth at the previous step W^(l) and the difference between the detected signal and the output of the neural network, i.e. Sdl^(l)-Y^<x) . The convergence rate η controls the convergence speed and the convergence stability. In fact, with a low convergence rate η , greater stability is obtained against slow learning by the neural network, whereas with a high η rate, a greater learning speed is obtained against a greater residual noise. The center μ-j_* of a gaussian function is iteratively updated. This is done by gradually moving this center over the entire interval comprised between the preceding center and the subsequent one and stopping the movement when the mean square error reaches the minimum value. In greater detail, if an iteration step is set to 0,05, at the beginning of the same iteration the center μ-_j* is set to μ-j_*-0,05 and the mean square error is evaluated.

The subsequent steps of iteration are as follows: 1) computation of the mean square error;

2) if the current mean square error is lower than that of the preceding step, the position of the center will be incremented by 0,05 and the algorithm will restart from the preceding step; 3) if the mean square error is greater than that of the preceding step, the center position will be decremented by 0,05 and the algorithm will end.

Thereby, the output of network 600 is optimized by updating both the weigths and the positions of center μ-j_* of a gaussian. Figure 7A is a diagram of a first waveform of an electroencephalograph^ potential EEG (relative to a single 125 ms recording sweep) before the latter undergoes the processing as described above. This signal has been obtained following the stimulation of the living creature's body during a surgery. Similarly, Figure 7B is a diagram of a second waveform relative to an evoked potential SEP reconstructed following the processing of signal EEG of Figure 7A. From the analysis of this second waveform information on the spinal cord functionality are obtained which are referable to the amplitude of evoked potential SEP and latency L, i.e. the time elapsing between the beginning of the recording sweep and the occurrence of a peak 700 of the waveform.

In fact, in the case that a cord distress is caused by surgical manoeuvres delaying the propagation of nervous pulses in the living creature's body being monitored, a concomitant delay of the occurrence of peak 700 and accordingly an increase in latency L will be observed.

As the latency L may vary between a sweep and the next one, in order to follow these variations the above processing method provides the adaptative variation of the center μ-_j* of the gaussian function located proximate to a negative peak 701 of the waveform of the registered signal EEG. Thereby, the result of the processing is the evoked potential SEP.

With relation to the operation of the second processing module 36, the latter operates to obtain from the samples of each sweep of signal S2 a maximum value for this signal and the instant of occurrence of such maximum value.

For example, the diagram of a further waveform relative to a H-reflex potential is illustrated in Figure 8. Useful information on monitoring derived from the analysis of the latter waveform concern the amplitude and a further latency L' relative to a further peak 800.

It should be noted that the results of the processing performed by modules 35 and 36 are sent to the system control unit 22 and from the latter to the display module 213. Particularly, the waveforms of the evoked potential SEP and the H-reflex being generated at one single electric stimulus can be displayed on the touch screen 13 of equipment 100.

It should be noted that, before being displayed, the results of the processing relative to SEP and H-reflex potential are converted from digital information to voltage levels that can be displayed. Numerical values indicating the maximum and minimum amplitudes of the bio-signals, as well as the respective latency values can also be displayed on this screen 13.

Alternative methodologies to the multi-layer neural network are known for the extraction of the evoked potential from one single recording. These methodologies provides the use of adaptative filters or other types of neural networks, frequently employed in applications other than monitoring.

FIR filters The method suggested by Chan F.H.Y., Lam F.K., Poon

P.W.F., W. Qiu ("Detection of brainstem auditory evoked potential by adaptive filtering", in Med. Biol. Eng.

Comput. , 1995), provides the use of FIR-type filters

(Finite Impulse Response) of the F-order with time- adaptable coefficients for the filtering of a reference input, which is strictly correlated with the waveform of the patient's evoked potential.

In this method, the reference signal is calculated by a recurrent synchronous average of the preceding recordings:

r_t=μ*r_t<+(l-μ)*x, (10) where xi and ri are the current recording and reference mean, respectively. As to the adaptation of the filter coefficients, it provides the minimization for each sweep of a figure of merit, typically the mean square error, as calculated based on the filter output and the current recording.

For the above method, the recording should be continuous, in order to avoid having to initialize the filter at each sweep.

ARX filter

The use of ARX filters, besides allowing the extraction of the single evoked potential, also allows a parametrisation thereof such as suggested by the method in Cerutti S., Baselli G., Liberati D., Pavesi G.,

"Single sweep analysis of visual evoked potentials through a model of parametric identification", in Biol. Cybernetics, 1987.

Particularly, it is assumed that the i^th recording Xi is given by the composition of the electroencephalographic activity, in this case being considered as the noise, and the evoked potential, i.e. the signal of interest; it is assumed that both these components are uncorrelated to each other.

x_i(k)=c_i(k)+n_i(k) (11)

The EEG signal of disturbance n_±(k) is modelled as an AR process driven by white noise e(k) whereas the single evoked contribution Ci(k) is estimated by filtering the reference signal ri(k), thus obtaining the following generation model:

ηOc-β+eβ) (12)

On the other hand, in frequency there is obtained: A(z)X=B(z)R+E (13) where

The value of the optimum triplet (n,m,d) is found by means of the minimization of FPE (Final Prediction Error) and AIC (Akaike Information Criterion) functionals.

Subsequently, the coefficients of ai and b_j are identified by the minimisation of the mean square error. Now, to obtain the estimate of the i-esimo evoked potential Ci, the reference ri is filtered by the coefficients ai and bj. In this method, the reference used is again a recurrent synchronous average of the preceding recordings.

Other types of neural networks

Multi-layer perceptron

An algorithm suggested by Dumitras A., Murgan A.T., Lazarescu V. ("A quantitative study of evoked potential estimation using a feedforward neural network" , in Proceedings of the 1994 IEEE Workshop on Neural Networks for Signal Processing, 1994) can be used, i.e. a three- layer neural network, one input, one hidden and one output layers, consisting of a different number of neurons. The function of activation of the hidden layer and the output layer is a sigmoid with the formula:

where α is a proportionality constant, ξ indicates the sigmoid slope and netl indicates the net input for that given neuron and will be

nlntpnt ^netIh = ∑^w//Λ ^{+ ι9}/ ⁽ 16 ⁾

1=1 for the neurons of the hidden layer and

nHidden netl _j = ∑w_hjx_h +S_h (17 )

/z=l for those of the output layer.

The error of each output neuron relative to the desired output for each example set being provided is calculated (with even more sophisticated norms than the

Euclidean one) and the weights among the various layers are updated through the backpropagation algorithm. To avoid the entrapment in local minima, variants of the basic algorithm are used, by introducing a momentum term, thus rendering the learning parameters adaptative according to analytical principles and exploiting second-order knowledge on error surface.

Wavelet-type neural networks.

In the method suggested by Heinrich H., Dickhaus H., Rothenberger A., Heinrich V., Moll G.H., "Single-sweep analysis of event-related potentials by wavelet networks - Methodological basis and clinical applications", in IEEE Transactions on Biomedical Engineering, 1999 the evoked potential of the i^th sweep is reconstructed by means of a neural network of the wavelet-type, i.e. by means a weighted mean of N wavelet functions:

Each wavelet function is characterised by 2 characteristic parameters: b_j, i.e. a translation; a-_j, i.e. a scaling. Particularly, a Morelet-type wavelet is used, thereby the estimation of y(t) takes place through the following formula:

y(t) = ( 19 )

The parameters a_j and α)j define the frequency of the node fj=ω-_j/2πa_j and are adapted in the course of learning. Each node of the network hidden layer is a time- frequency characteristic of the evoked potential. To avoid that during the learning the nodes may compete for the higher-energy components, they shoould be trained separately and recursively.

At the j^th stage of recursion the j^th neuron is trained based on the residue βj (t) , i.e. the difference between the recording and the sum of outputs of the already trained neurons. This procedure is repeated until a preset constraint is respected, typically concerning the signal energy represented by the network.

In order to obtain the weights and parameters of the network, one starts by training a network identical to the subject one, but with a synchronous mean of evoked potentials rather than with one single recording. Thereby, a scaling, a translation and a reference frequency, i.e. a_ref,j, b_ref,j ^e ω_ref,j, are obtained per each neuron. Starting from them, a bandpass filter and a time window are obtained to be used for the corresponding neuron in the network which should extract the single- sweep potential, such that it can learn only specific time-frequency characteristics of the recording, the same characteristics being analysed by the equivalent thereof in the network trained with synchronous mean.

Thereby, a signal ej (i.e. ej filtered and limited over time) is obtained and the mean square error between §j and the output of the j^th neuron is minimised. Subsequently to the separate training of each neuron, a final learning simultaneous for all neurons (though retaining the bandpass filters and the time window) is performed.

Hybrid methods

Neural network and FIR or ARX filter

Since better performances are obtained with filters with a reference input having a high correlation with the signal to be extracted, the output of the inventive neural network can be used as the reference for a FIR or

ARX filter.

NARX (non-linear ARX) by means of neural network In order to prevent that the non-linear component inherent in the signal to be extracted may be lost, a non-linear ARX filter may be applied, having the formula:

y(k)=∑a(i)u(k-i)+∑b(j)y(k-j)+ i=o J=\

P P +∑∑a(i,j)u(k-i)u(k-j)+∑∑b(i,j)y(k.i)y(k-j)+ i=Q J=O 1=1 j=\ P Q

+∑∑c(ij)u(Ic-i)y(k-j)+...+e(k) (20)

/=o j=\

In a matrix form, the preceding expression can be summarized as follows:

y(k)=a^Tu+b^Ty+u^TAu+y^TBy+u^TCy+...+e(k) (21)

Considering the matrix formulation of the input- output relationship in a general network y(k)=H^TG(x)+e(k) (22) where

G(x)

giOh) - SN(^XN)J (23) and

(with n=l..JST) and expanding the activation functions into Taylor series, one obtains

A h₂g₂₂w₂wζ + ... + h_Ng_2Nw_Nw_N ^T (26a)

B ₂g₂₂v₂vζ + ... + h_Ng_2Nv_Nv_N ^T (26b)

C

+ h₂g₂₂w₂vζ + ... + h_Ng_2Nw_Nv_N ^T) (26c) where g_tn indicates the t derivated of the activation function of the n^th neuron such as described by Wang D., Lum K. -Y., Yang G. in the document "Parameter estimation of ARX/NARX model: a neural network based method" in Proceedings of the 9th International Conference on Neural Information Processing, 2002.

In this way, the NARX parametres are identified, which can then be used for filtering the reference signal, thereby obtaining an estimate of the single-sweep evoked potential. The selected activation functions are of the polynomial, exponential or sigmoid types.

Neural network and non-linear filter

The output of the neural network is used to supply the non-linear filter, as also in this case the performance is improved if the reference signal is more correlated with the signal to be extracted.

It should be observed that with equipment 100 and method of the invention it is possible to compare the morphological variations of the bio-signal waveforms in the basic components thereof (amplitude and latency variations) thus providing indications that can be correlated with the functionality status of the spinal cord. In the case where both signals are significantly and simultaneously modified following a surgical manoeuvre, a status of cord distress can be reliably supposed. For example, in the case of serious cord damages, both signals disappear from the screen 13.

On the contrary, with the modification of only one of both bio-signals, cord distress may not be as likely.

In addition, by a rea—time comparison of the variability of the H-reflex to that of SEP potential the cord functionality can be inspected in its anatomical wholeness by a non-invasive method which increases the monitoring reliability.

Furthermore, equipment 100 allows to extract the SEP evoked potential from the electroencephalographic potential EEG being detected following a single electric stimulus without having to average the EEG potentials detected in correspondence of a number of stimuli. This equipment 100 integrates both the stimulation and the recording and processing functions within only one compact structure.

Finally, the inventive equipment 100 can be used also by operators who are not specialised in the neurophisiology field due to the easy positioning of the electrodes in standardized positions and to the visualization of the bio-signal waveforms. In a variant embodiment of the monitoring method, it can be envisaged to apply the electric stimulus to two different stimulation sites, i.e. one per each leg in a non-simultaneous way. In this case, two distinct groups of stimulation electrodes per each leg and two distinct groups of recording electrodes for the H-reflex potential should be used. Thereby, it would be possible to extract the H-reflex from each leg to be compared with the corresponding SEP potential. This may be carried out either by using two pieces of equipment such that the inventive one can be operated alternatively or by means of a dedicated equipment.

Obviously, those skilled in the art, aiming at satisfying specific and contingent requirements, may carry out further modifications and variants both to the equipment and the monitoring method of the present invention, all being contemplated within the scope of protection of the invention, such as defined by the following claims.

Claims

Claims

1. A method for monitoring the functionality of the spinal cord of a living creature, comprising the steps of: a) acquiring an electric signal (Sl) obtained in response to an electric stimulation applied to at least a first stimulation site of the creature's body and comprising a first signal (S) indicative of a somatosensory evoked potential generated by said stimulation; b) acquiring a second electric signal (S2) indicative of a H-reflex potential generated by said electric stimulation; c) processing the first and second electric signals to simultaneously provide first (L) and second

(L' ) information, which are correlated to the first and second electric signals, thus enabling a combined detection of the functionality of the spinal cord.

2. The monitoring method according to claim 1, wherein said electric signal (Sl) is indicative of an encephalographic potential comprising a noise component (Ns) and said first electric signal (S) , said step c) comprising a further step of processing the electric signal (Sl) to extract said first electric signal (S) .

3. The monitoring method according to claim 1, comprising the steps of: acquiring a test electric signal (Sl'; S2' ) from the living creature indicative of a bio-signal independent from the application of the electric stimulation; analysing at least one characteristic of the test signal (Sl'; S2' ) to check for the correspondence of the same with a preset evaluation principle; enabling the delivery of the electric stimulation to acquire the electric signal (Sl) and the second electric signal (S2) in consequence of the correspondence of said at least one characteristic with the evaluation principle and the inhibition thereof in consequence of the non-correspondence with the principle.

4. The monitoring method according to claim 3, wherein the test signal (Sl') is indicative of the electroencephalograph^ potential bio-signal and the preset evaluation principle of said at least one characteristic of the test signal comprises the steps of : - analysing the maximum value and minimum value of the test signal; analysing an amplitude of low-frequency oscillations of the signal.

5. The monitoring method according to claim 4, wherein the preset evaluation principle of said at least one characteristic of the test signal (Sl') comprises the further steps of: checking the occurrence of quick oscillations of the signal; - analysing the amplitude of the high-frequency components of the signal.

6. The monitoring method according to claim 3, wherein the test signal (S2') is indicative of the electromiographic muscular potential bio-signal and the preset evaluation principle of said at least one characteristic of the test signal (S2') comprises the step of analysing the maximum value and the minimum value of the test signal.

7. The monitoring method according to claim 1, wherein said step of acquiring said electric signal (Sl) and said second electric signal (S2) comprises the steps of: detecting said signal and said second signal in correspondence of at least one first (514) and at least one second (517) recording sites, respectively; carry out on said signal and on said second signal a high-pass analogue filtering and a low-pass analogue filtering to eliminate non-significant signal frequency components; - amplifying said signals with different amplification gains to optimize the peak amplitude; carry out an analogue/digital conversion of the signals thus obtaining quantized samples to be processed, which are contained in a recording time interval of a preset amplitude.

8. The monitoring method according to claims 2 and 7, wherein said further step of processing the electric signal (Sl) comprises the steps of: d) generating an estimate signal (Y) of the first electric signal (S) , based on at least one recording time interval containing the signal samples, ; e) comparing said estimate signal (Y) to the electric signal (Sl) ; f) updating the estimate signal (Y) to minimize an error between the latter and the first signal (S) ; g) repeating the steps d) , e) and f) with another recording time interval.

9. The monitoring method according to claim 8, wherein said step of generating the estimate signal is carried out in agreement with a logic based on a neural network.

10. The monitoring method according to claim 9, wherein said neural network is a radial base multi-layer neural network.

11. The monitoring method according to claim 10, wherein said multi-layer neural network substantially operates, in accordance to the expression:

where X is a input vector to the neural network,

Y(X) is the estimated value of the first electric signal

(S) , μ_j and σ² are the mean and the variance of a function of the Gaussian-type associated to a j^th hidden neuron of the network and w^(l) _j is the weight attributed to the output of the same neuron for the i^th recording time interval.

12. The monitoring method according to claim 11, wherein said step of updating the estimate signal (Y) comprises the further steps of: - updating the weights w^(l) with an algorithm of the least mean squares (LMS) substantially based on the following relation in the matrix form for the i^th time interval: _w ⁽ⁱ ₊ ⁱ⁾ _{= w} ⁽ⁱ⁾ ₊₂η_H(Sdl⁽ⁱ⁾-Y⁽ⁱ⁾) where w^(l+1> indicates the updated weight, W^(l) indicates the weight of the preceding step, Sdl^(l)-Y^(l) is the difference between the acquired electric signal and the estimate signal and η is a convergence rate; iteratively updating the mean μ-j* of at least one Gaussian-type function by moving the latter on the interval comprised between the preceding mean and the subsequent one.

13. The monitoring method according to claim 1, further comprising a visualization step of a first and second waveforms relative to the first (S) and second

(S2) electric signals, respectively, in order to obtaiin the first and second information on the spinal cord functionality.

14. Equipment (100) for monitoring the spinal cord functionality in a living creature, comprising: a first (27) acquisition module to acquire an electric signal (Sl) obtained in response to an electrical stimulation and comprising a first electric signal (S) indicative of a somatosensory evoked potential generated by said electric stimulation; a second (28) acquisition module to acquire a second electric signal (S2) indicative of a H-reflex potential generated by said electric stimulation; a stimulation module (24) to apply said electric stimulation to a first stimulation site in the living creature's body; a main block (22) to manage the processing of the first and second electric signals and to simultaneously provide first (L) and second (L' ) information correlated with the first and second electric signals thus allowing a detection of the spinal cord functionality.

15. Equipment (100) according to claim 14, wherein said electric signal (Sl) is indicative of an electroencephalograph^ potential comprising a noise component (Ns) and said first electric signal (S) , said equipment comprising a first processing module (35) of the electric signal (Sl) to extract said first electric signal (S) .

16. Equipment (100) according to claim 14, further comprising an evaluation module (33; 34) of a test electric signal (Sl'; S2') acquired by the living creature and indicative of a bio-signal independent from the application of the electric stimulation, said evaluation module being able to verify the correspondence of said test signal to a preset evaluation principle to send to the main block (22) instructions (31'; 32') to enable or inhibit the delivery of the electric stimulation based on this verification.

17. Equipment (100) according to claim 16, wherein said evaluation module (33; 34) is provided by logical blocks of a program being implemented by the main block

(22) or by an electronic device.

18. Equipment (100) according to claim 15, wherein said first processing module (35) of signal (Sl) comprises logical blocks of a program implemented by the main block (22) .

19. Equipment (100) according to claim 15, wherein said first (27) and second (28) acquisition modules are electronic devices comprising circuits to receive and filter the signal (Sl) and the second signal (S2) , said circuits having parameters that can be modified by the main block (22) through numerical control instructions (25, 26) .

20. Equipment (100) according to claim 14, wherein said stimulation module (24) is an electronic device comprising digital and analogue circuits and having an output stage (44, 45) to send a voltage or current electrical stimulation.

21. Equipment (100) according to claim 14, wherein a memory module (40) is connected to the main block (22) to store the first and second information correlated to the first (S) and second (S2) electric signals, respectively.

22. Equipment (100) according to claim 14, comprising a display module (213) connected to the main block (22) to display a first and second waveforms relative to the first (S) and second (S2) electric signals, respectively.

23. Equipment (100) according to claim 22, wherein said display module (213) includes a touch screen (13) to help the quick transmission of external commands (42) to the main block (22) .

24. Equipment (100) according to claim 14, wherein said electric stimulation may be a single electric pulse, a train of electric pulses, or rather biphasic electric pulses.