EP0938685A1

EP0938685A1 - Method and device for processing signals from a set of photodetectors with cellular architecture, and application to gamma-cameras

Info

Publication number: EP0938685A1
Application number: EP97913263A
Authority: EP
Inventors: Edmond Tournier; Bernard Chabert; Alain Chapuis
Original assignee: Commissariat a lEnergie Atomique CEA
Current assignee: Commissariat a lEnergie Atomique et aux Energies Alternatives CEA
Priority date: 1996-11-08
Filing date: 1997-11-06
Publication date: 1999-09-01
Also published as: FR2755816B1; JP2001503526A; FR2755816A1; WO1998021606A1

Abstract

The invention concerns a method for processing signals produced by a set of N photodetectors, in response to a phenomenon to be detected, in which a cellular architecture (88, 90-1....90-6, 92, 94), associated with each photodetector (60), enables to compare the flux it receives with the flux of the neighbouring photodetectors (160-1,...,160-6), and to determine which set of photdetectors to take into account for a given interaction. Characteristic quantities of the interaction can then be computed.

Description

METHOD AND DEVICE FOR PROCESSING SIGNALS OF A

SET OF PHOTODETECTORS HAVING ARCHITECTURE

CELLULAR, AND APPLICATION TO GAMMA-CAMERAS

DESCRIPTION

Technical area

The present invention relates to the processing of data or signals originating from, or transmitted by, a set of N photodetectors, in response to an event to be characterized or identified, for example to locate.

The invention applies in particular to the determination of the characteristics, for example of energy and of position, of an event from signals supplied by photomultipliers equipping a gamma-camera; for example, the position of an event is identified with respect to the photomultipliers themselves. By gamma-camera is meant a camera sensitive to gamma radiation (γ). Such cameras are used in particular for medical imaging purposes.

State of the art

At present, most gamma cameras used in nuclear medicine are cameras operating on the principle of Anger type cameras. We can refer to this subject in the document

US-3,011,057.

Gamma cameras allow in particular to visualize the distribution, in an organ, of molecules marked by a radioactive isotope previously injected into the patient. The structure and operation of a known gamma camera are described and summarized below with reference to Figures 1, 2A and 2B attached.

Figure 1 shows a detection head 10 of a gamma camera placed in front of a body 12 containing molecules isotopically labeled radio ^¬ active.

The detection head 10 includes a collimator

20, a scintillator crystal 22, a light guide 24 and a plurality of photomultiplier tubes 26 juxtaposed so as to cover one face of the light guide 24. The scintillator is, for example, a Nal crystal (T-f?).

The function of the collimator 20 is to select from all the gamma rays 30 emitted by the member 12 those which reach the detection head substantially under normal incidence. The selective nature of the collimator makes it possible to increase the resolution and the sharpness of the image produced. However, the increase in resolution comes at the expense of sensitivity. For example, for approximately 10,000 γ photons emitted by the organ 12, a single photon is effectively detected.

The γ photons having passed through the collimator reach the scintillator crystal 22 where almost each γ photon is converted into a plurality of light photons. In the rest of the text, an event designates each interaction of a gamma photon with the crystal, causing a scintillation. The photomultipliers 26 are designed to emit an electrical pulse proportional to the number of light photons received from the scintillator for each event. So that a scintillation event can be located more precisely, the photomultipliers 26 are not directly attached to the scintillator crystal 22 but are separated from the latter by the light guide 24.

Photomultipliers emit a signal whose amplitude is proportional to the total quantity of light produced in the scintillator by gamma radiation, that is to say proportional to its energy. However, the individual signal of each photomultiplier also depends on the distance which separates it from the point of interaction 30 of the gamma radiation with the material of the scintillator. Indeed, each photomultiplier delivers a current pulse proportional to the light flux it has received. In the example of FIG. 1, small graphs A, B, C show that photomultipliers 26a, 26b and 26c located at different distances from an interaction point 30 deliver signals with different amplitudes.

The position of the interaction point 30 of a gamma photon is calculated in the gamma camera from the signals coming from the set of photomultipliers by carrying out a barycentric weighting of the contributions of each photomultiplier.

The principle of barycentric weighting as it is implemented in cameras of the Anger type appears more clearly by referring to FIGS. 2A and 2B appended.

FIG. 2A shows the electrical wiring of a detection head 10 of a gamma camera, which connects this camera to an imaging unit. The detection head comprises a plurality of photomultipliers 26.

As shown in FIG. 2B, each photomultiplier 26 of the detection head is associated with four resistors denoted RX ^~ , RX ⁺ , RY ^" and RY ⁺ . The values of these resistors are specific to each photomultiplier and depend on the position of the photomultiplier in the detection head 10.

The resistors RX ^~ , RX ⁺ , RY ^" and RY ⁺ of each photomultiplier are connected to the output 50 of said photomultiplier, represented in FIG. 2B with a symbol of current generator. They are on the other hand respectively connected to collector lines common noted LX ^~ , LX ⁺ , LY ^" , LY ⁺ , in Figure 2A. The lines LX ^~ , LX ⁺ , LY ^" and LY ⁺ are in turn connected respectively to analog integrators 52X ^" , 52X ⁺ , 52Y ^" , 52Y ⁺ , and, through these to analog converters / digital 54X ^" , 54X ⁺ , 54Y ^" , 54Y ⁺ . The output of the converters 54X ^" , 54X ⁺ , 54Y ^" , 54Y ⁺ is directed to a digital operator 56. The lines LX ^" , LX ⁺ , LY ^" , LY ⁺ are also connected to a common channel, called the energy channel, which also includes an integrator 57 and an analog / digital converter 58 and its output is also directed to the operator 56.

Thanks to the device of FIG. 2, the position of the interaction is calculated according to the following equations (US-4,672,462): _ x + -χ- x ++ χ-

and

_v _ Y + -Y- Y ++ Y-

in which X and Y indicate the coordinates along two orthogonal directions of the position of the interaction on the crystal and in which X ⁺ , X ^" , Y ⁺ , Y ^" indicate respectively the weighted signals delivered by the integrators 52X ⁺ , 52X ^" , 52Y ⁺ , 52Y ^" .

The values of X and Y as well as the total energy E of the gamma ray having interacted with the crystal are established by the digital operator 56. These values are then used for the construction of an image as described, for example, in document FR-2,669,439.

The calculation of the position of the interaction is marred by an uncertainty linked to the Poisson statistical fluctuations of the number of light photons and the number of photoelectrons produced for each event, that is to say for each gamma photon detected. The standard deviation of the fluctuation is lower the higher the number of photons or photoelectrons. Because of this phenomenon, light should be collected as carefully as possible. The intrinsic spatial resolution of the camera is characterized by the width at half-height of the distribution of the positions calculated for the same collimated point source placed on the scintillator crystal. For gamma rays with an energy of 140 keV, the resolution is generally of the order of 3 to 4 mm.

The energy of a detected gamma photon is calculated by summing the contributions of all the photomultipliers that have received light. It is also marred by a statistical fluctuation. The energy resolution of the camera is characterized by the ratio of the width at half-height of the distribution of the energies calculated to the average value of the distribution, for the same source.

The energy resolution is generally of the order of 9 to 11% for gamma rays with an energy of 140 keV.

Finally, an Anger type gamma camera has the advantage of making it possible to calculate in real time the barycenter of the photomultiplier signals with very simple means.

Indeed, the system described above comprises a limited number of components. In addition, the resistors used to inject the signal from the photomultipliers into the collector lines are very inexpensive.

However, such a camera also has a major disadvantage which is a reduced counting rate. The counting rate is understood to mean the number of events, that is to say interactions between a γ photon and the scintillator, which the camera is capable of processing per unit of time.

One of the limitations of the counting rate comes in particular from the fact that the camera is unable to process two events taking place substantially simultaneously at points distinct from the scintillator crystal. Indeed, simultaneous but geometrically distinct events give rise to electrical signals which pile up in the collector lines LX ^" , LX ⁺ , LY ^" and LY ⁺ and which can no longer be distinguished. These events are also "lost" for imaging.

Gamma cameras are used not only in traditional imaging techniques, but also in two other medical imaging techniques where limiting the counting rate is also a crippling constraint.

These techniques are the techniques known as “attenuation correction by transmission” and “PET (Positon Emission Tomography) in coincidence”. The transmission attenuation correction technique consists in taking into account, when forming a medical image, the proper attenuation of the patient's tissue surrounding the organ under examination. To find out this attenuation, the transmission of gamma radiation to a gamma camera is measured through the patient's body. For this purpose, the patient is placed between a very active external source and the detection head of the gamma-camera. Thus, when measuring the transmitted radiation, a high number of events take place in the scintillator crystal. The high number of events per unit of time also increases the probability of having several substantially simultaneous events. A classic Anger type camera is therefore inappropriate. The PET technique consists in injecting the patient with an element such as F ¹⁸ capable of emitting positrons. The annihilation of a positron and an electron releases two γ photons emitted in opposite directions and having an energy of 511 keV. This physical phenomenon is exploited in the PET imaging technique. In this technique, a gamma camera is used with at least two detection heads arranged on either side of the patient. The detection heads used are not equipped with a collimator. Indeed, electronic processing of information, known as coincidence processing, makes it possible to select from among the events those which coincide in time, and thus to calculate the trajectory of the gamma photons.

The detection heads are therefore subjected to high fluxes of gamma radiation. Conventional Anger type gamma cameras have a counting rate which is generally too limited for such an application. As an indication, a gamma-camera type

Anger can operate normally with detection of 1.10 ⁵ events per second, while in PET imaging it takes at least 1.10 ⁶ events per second for normal operation. Another limitation of gamma cameras of the type

Anger, described above, is due to the fact that the calculation of the barycenter of an event is definitively fixed by the construction of the detection head and in particular by the choice of resistors RX ^" , RX ⁺ , RY ^" , RY ⁺ for each photomultiplier. Similarly, the energy calculation is fixed by the wiring of the photomultipliers on a common channel (energy channel).

A conventional device therefore attributes an event to a photodetector and only one, and takes into account all the signals of all the photodetectors for the calculation of the energy. However, the photomultiplier amplifiers not affected by an event add noise to the LX ⁺ , LX ^" , LY ⁺ , LY ^" collector lines.

STATEMENT OF THE INVENTION The subject of the invention is a method and a device in which only the information, or the signals, originating from the Ni photodetectors concerned, from a set of N photodetectors, by an event, is digitized and undergoes processing local to isolate the maximum intensity signal, or maximum amplitude, and a number of signals emitted by the photodetectors near the photodetector having emitted the maximum intensity signal, or maximum amplitude.

More specifically, the subject of the invention is a method for processing signals produced by a set of N photodetectors in response to an event to be detected, comprising:

- comparison, with a threshold value, for each photodetector, of the intensity or of a digital integral of the digitized signal obtained from the transmitted signal,

- the comparison, for each photodetector emitting a signal whose intensity or the digital integral exceeds said threshold, of the intensity or the integral of the signal which it emits with the intensity or the integral of the signal emitted with each of the neighboring photodetectors,

- memorizing the value of the maximum intensity or integral among all the intensities or the integrals compared, as well as the intensities or integrals of the signals emitted by the photodetectors near the photodetector whose signal has the intensity or l '' maximum integral. It is then possible to produce, using the stored intensities, at least one signal representative of a quantity characteristic of an event produced in front of the photodetectors. Thus, only the signals from the photodetectors concerned by the interaction are stored or contribute to the calculation, or to the production of a representative signal, of a quantity characteristic of the event. It is not necessary to memorize or take into account the signals produced by all the photodetectors, since the reading time would cause a sharp deterioration in the count rate performance. Each event concerns in principle only a small number of photodetectors, the method makes it possible to retain only those photodetectors which are affected by the event. In addition, two events occurring simultaneously, or almost simultaneously, in two different places, relative to the set of N photodetectors can then be taken into account. The invention also relates to a method for identifying a characteristic of an event occurring before a set of N photodetectors, comprising:

- the identification of a subset of Ni photodetectors, concerned by the interaction, that is to say having produced a signal in response to the event,

the production of a signal representative of the characteristic of the event, as a function: * of the signal of maximum intensity or maximum digital integral among the N _x signals produced by the i photodetectors, in response to the event, * and signals emitted by the photodetectors close to the photodetector whose signal has the maximum intensity or the maximum digital integral among the Ni photodetectors. This second method makes it possible to generate signals representative of a characteristic of an event occurring in front of the set of N photodetectors, solely as a function of the signals emitted by the photodetectors concerned by said event. Here again, an unnecessary reading of the entire field of N photodetectors and of the corresponding memories is avoided, a reading which would consume a great deal of time and which would reduce the counting rate.

In both cases, a storage device is associated with each of the N photodetectors. It is in the memory associated with the photodetector having emitted the maximum intensity signal that this maximum intensity and the intensities, or the digital integrals, of the signals coming from the photodetectors near the photodetector whose signal has the maximum integral, are stored.

The subject of the invention is also devices making it possible to implement the above methods. In particular, the subject of the invention is a device for processing signals produced by a set of N photodetectors, a storage device being associated with each of the N photodetectors, this device comprising: means for comparing, for each photodetector, the intensity of a signal transmitted with a threshold value, - means for comparing, for each photodetector emitting a signal whose intensity exceeds said threshold, the intensity of the signal it emits with the intensity of a signal emitted by each of the neighboring photodetectors,

means for storing, in the storage device associated with the photodetector emitting the maximum intensity signal, the value of the maximum intensity among all the intensities, as well as the intensities of the signals emitted by the photodetectors near the photodetector, the signal of which presents the maximum intensity.

A subject of the invention is also a device for identifying a quantity characteristic of an event occurring before a set of N photodetectors, a storage device being associated with each of the N photodetectors, this device comprising:

- means for identifying a subset of Ni photodetectors, having produced a signal in response to the event, - means for storing, in the storage device associated with the photodetector emitting the maximum intensity signal, the value of l maximum intensity among all the intensities, as well as the intensities of the signals emitted by the photodetectors close to the photodetector whose signal has the maximum intensity,

means for producing a signal representative of the quantity characteristic of the event, as a function of the intensities: * of the signal of maximum intensity among the Ni signals, produced by the N _x photodetectors in response to the event, * signals emitted by neighboring photodetectors, called first ring, of the photodetector whose signal has 1 intensity among the Ni photodetectors. The invention has been described using the digital integral as a signal representative of the intensity of each signal coming from each photodetector.

In fact, it is also possible to use, instead of the digital integral, the maximum intensity of the digital signal or of the analog signal. However,

1 digital integral is the information which is best suited to the implementation of the invention.

Brief description of the figures

In any case, the characteristics and advantages of the invention will appear better in the light of the description which follows. This description relates to the exemplary embodiments, given by way of explanation and without limitation, with reference to the appended drawings in which:

- Figure 1, already described, is a schematic section of a detection head of a gamma-camera, for example of the Anger type; - Figures 2A and 2B, already described, schematically show a device of the prior art, for the collection and coding of signals from photomultipliers of the detection head of Figure 1; - Figures 3A and 3B show two examples of photodetector fields, with different symmetries; - Figures 4, 5A and 5B show a part of a device associated with a photodetector;

- Figures 6A and 6B show an analog signal supplied by a photodetector (Figure 6A), as well as the corresponding digital signal (Figure 6B);

- Figure 7 shows an embodiment of one invention;

FIG. 8 schematically represents a reference used for the calculation of the local coordinates of an event with respect to a field of photodetectors;

- Figures 9A and 9B show photodetector signals in a case of simple event and in a case of stacking;

FIG. 10 is a timing diagram for signals transmitted in a device according to the invention,

- Figure 11 is another embodiment of one invention; - Figures 12 and 13 show embodiments of functional elements in a device according to one invention;

- Figures 14 and 15 show schematically embodiments of components for a device according to the invention.

Detailed description of embodiments of the invention

The example will be taken from a set of N photodetectors, or photomultipliers, from a gamma-camera. The number N is then substantially between 50 and 100. Such a set of N photodetectors is preferably a two-dimensional network of photodetectors, for example a two-dimensional network of photomultipliers. Examples of two-dimensional arrays of photodetectors are illustrated in FIGS. 3A and 3B. More precisely, these figures represent, in each case, a position of a network of photodetectors in plan view (or, in other words, if one refers to FIG. 1, seen from the scintillator 22) . The cross section of a photodetector can have various shapes, for example square, hexagonal or circular.

FIG. 3A represents, in top view, a mixed field of hexagonal and round photodetectors. The hexagonal photodetectors are arranged in a honeycomb pattern.

FIG. 3B corresponds to a field of photodetectors, seen from above, each photodetector having a square section.

When an event occurs in front of the photodetector field, this event emitting radiation in a spectral range corresponding to that of the photodetectors, a certain number of photodetectors will emit signals. In FIG. 3A, the sets of photodetectors concerned, that is to say emitting signals, have been represented for two events occurring at different locations relative to the photodetector field. A first event occurs in front of the photodetector 60 and the latter produces a signal. In response to this event, the neighboring photodetectors of photodetector 60 also produce a signal: these are, in FIG. 3A, photodetectors 160 of the first ring (in moderately tight hatched lines), immediately adjacent to the photodetector 60, and of the photodetectors 260 (in spaced hatched lines), said to be of the second ring around the photodetector 60. Beyond, the signals emitted by the other photodetectors are generally negligible. Likewise, an event occurring opposite the photodetector 61 results in the emission of signals by the photodetector 61, but also by the photodetectors of its first ring 161 (in medium tight hatched lines) and by the photodetectors 261 of its second ring (in lines hatched spaced).

In FIG. 3B, an event occurring opposite a photodetector 63 also produces a signal in the neighboring photodetectors 163 (first ring) and in the photodetectors 263 (second ring) around the photodetector 63.

Figure 4 is a partial view of a signal processing device of a photodetector. It represents only a single channel of such a device, that is to say the part of the device associated with a single photodetector 60. Similar channels are associated with the other photodetectors, in particular the first and second crown photodetectors which surround the photodetector 60. The photodetector 60 is connected to a current-voltage converter 62. In response to a detected event, a signal is obtained on the output 64 of the converter 62, for example of the type illustrated in the figure 6A.

The graph in FIG. 6A indicates, on the ordinate, the amplitude of the signal corresponding to impulse and, on the x-axis, time. The amplitude of the signal and the time are indicated in arbitrary scales. t ₀ designates the starting instant of the pulse supplied by the photodetector, and ti the instant when the pulse becomes almost zero again, after having passed a maximum. As an indication, the duration corresponding to the interval ti - t ₀ is of the order of a microsecond, in the case of a photomultiplier of a gamma-camera coupled to a Nal crystal (Υl).

The analog signal present on the output terminal 64 is directed to an analog-digital converter 66. The latter samples each pulse of the signal in a number of samples n, as illustrated in FIG. 6B. Two consecutive samples are separated by a step, or clock interval p (the clock operating at 1 / p Hz).

As an example, the converter samples each pulse of the signal in n = 10 samples. For a signal of 1 microsecond, a sampling is then carried out every 100 nanoseconds.

The analog-digital converter 66 is preferably a fast converter, of the “flash” type which can operate at a frequency of the order of 10 to 20 megahertz.

The digital signal from the analog-digital converter 66 directed to a digital integrator 68 (composed of a shift register with N flip-flops and a summator). This integrator performs a sliding sum of the samples transmitted to it by the analog-digital converter 66. The sliding sum is performed on a given number of samples. This predetermined number is, for example, 10.

The signal representative of the value of the coded signal (at the output of 66) is sent to means 70 capable of detecting the presence of an event.

FIG. 5A represents an embodiment of the detection of events by detecting the passage of the signal coded by a maximum and by validating it when this signal is above a predetermined threshold. The output of the digital analog encoder 66 is connected to one of the inputs of a comparator 702. To the second input 701 is applied a predetermined digital value VP. This comparator outputs a signal 703 equal to 0 if the coded value is less than VP and 1 if the coded value is greater than VP. Signal 703 is connected to one of the inputs of an ET 704.

The coded output is also connected to one of the inputs 705 of a second comparator 708 and to the input of the register 706 which delays the coded value by a clock step. The output 707 of this register is connected to the second input 707 of the comparator 708. This comparator emits a pulse 709 equal to 1 when it sees for the first time the input 707 (which represents the value coded at the instant t- 1) greater than input 705 (which represents the value coded at time t). The output 709 is equal to 0 in the other cases. This output 709 is connected to the second input of the AND circuit, 704. The output signal 710 of the AND represents the instant of passage of the coded signal by a maximum when the coded signal is greater than VP. Signal 710 is connected to the input of an offset circuit 712 which delays it by a certain number of steps clock. This number is adjusted so that the pulse at its output is synchronous from the instant when the output value of the sliding adder 68 is equal to the integral of the detected signal. FIG. 5B represents an embodiment of the sliding summator 68 which performs the integration of the event. The coded signal at the output of the analog-digital converter is directed to a shift register 682 whose number of stages represents the duration of integration. The output of the shift register 682 is connected to the subtracting input of a subtractor 684 while the positive input is connected to the coded signal. The output of the subtractor is sent to an accumulator 686. The output of the accumulator 686 represents the integral of the event when the circuit 712 delivers its pulse.

The set of means, associated with a photodetector, and described above in conjunction with FIG. 4 is designated by the reference 88. Each set of means is associated with each photodetector.

Each photodetector then compares the value of its digital integral with that of the neighboring photodetectors. A device for carrying out this comparison is shown in FIG. 7. In this figure, the neighboring photodetectors, called first ring photodetectors, are represented schematically by the references 160-1, ..., 160-6. Each of these photodetectors is associated with a set of means equivalent to the means 88 associated with the photodetector 60. The digital integral corresponding to the signal delivered by the photodetector 60 is compared with the digital integral of the signal delivered by each of the photodetectors 160-1, ..., 160-6, and this using comparison means, or comparators 90-1, ..., 90-6. The comparators work continuously (at each coding step) and provide a comparison result which takes the value 1 when the value of the current integral of the photodetector 60 is greater than the current integral of the neighboring photodetector to which it is compared, and the value 0 otherwise. The 6 comparison results are sent to a device 92 (FIG. 7) composed of an AND circuit with 6 inputs, which delivers a signal equal to 1 when all the comparison results are equal to 1, that is to say when the value of the current integral of the photodetector 60 is greater than that of all of its neighbors. The output of the AND circuit is a first condition which makes it possible to validate, or not, the signal EVT (or DMDS storage request) supplied by the circuit 712 (FIG. 5A), and which results from the detection of a maximum on the signal coded. The digital integral corresponding to the signal delivered by the photodetector 60 (that is to say the maximum intensity signal) and the digital integrals corresponding to the signals delivered by the photodetectors 160-1, •• -, 160-6 ( that is to say the signals from the photodetectors of the first ring) are stored in the storage device 94 associated with the photodetector 60.

Once these values have been memorized, means, which can be integrated into the assembly 94, can make it possible to calculate any quantity characteristic of the event. Thus, it is possible to calculate the total energy of the event: for this, the sum 6 E = ∑ Ej_ of all the digital integrals E _x i = 0 corresponding to the main photodetector 60 and to the neighboring photodetectors 160-1, ..., 160-6 is carried out.

It is also possible to calculate, or estimate, the local coordinates of the event, relative to the field of N photodetectors, for example by formulas of the type:

6

Σ EiXid

X = ≡ O,, v

6

∑Ei i = 0

Σ EiY ^

Y = i = 0 6

∑Ei i = 0

or :

- d represents the distance separating two adjacent photodetectors on the X axis (between the centers of these two photodetectors),

1 is the distance separating two adjacent photodetectors on the Y axis,

- and where the coordinates of photodetectors 160-1, ..., 160-6 are identified by their center in the form (X _lf Y with Xiel-1,0,1} and 0, 1}, in an orthogonal coordinate system (0, i, j) as illustrated in FIG. 8, where 0 represents the middle of the central photodetector 60, i the unit vector along the X axis, and j the vector unitary along the Y axis. The characteristic magnitude of the event, once calculated, can then be transmitted to the outside.

In addition to the comparisons carried out using the comparators 90-1, ..., 90-6, the value of the digital integral of the photodetector 60 is transmitted to a set of comparators, each of which is associated with one of the peripheral photodetectors 160 -1, ..., 160-6, the photodetector 60 then being treated as a photodetector of the first ring for the other photodetectors. Thus, each photodetector of the entire field of N photodetectors is associated with a device of the type which is illustrated above in FIG. 7, including a storage device 94 and, optionally, a calculation device.

Because of the comparisons made in the comparators associated with each photodetector, that is to say because of the comparisons made locally, it is indeed the photodetector producing a signal, of intensity or of energy, maximum which is considered as "central" photodetector for the event, its neighboring photodetectors then being considered as first ring photodetectors for the event; Similarly, it is then the digital integral of this central photodetector which is stored or memorized as a "main" signal, while the digital integrals corresponding to the photodetectors of the first ring are stored or memorized in the other registers, and correspond to secondary values, each being less than the value of the digital integral associated with the central or main photodetector. The methods and devices described above are those of a "cellular architecture", which allows photodetectors to compare the light fluxes they receive with each other, and to determine which of them are affected by the interaction. The storage of the maximum intensity and the intensities of the neighbors of the first ring can therefore be carried out in the storage device associated with the central photodetector, that is to say with the photodetector which delivers the maximum intensity signal. Such a method makes it possible to improve the performance of the counting rate of the photodetectors, thanks to the local processing of digitized information.

In addition, two events occurring simultaneously, in two places sufficiently distinct from each other, can be taken into account. Again, this increases the count rate.

With regard to the calculation of the characteristic (s) of the event, once the central photodetector and the neighboring photodetectors have been identified, two possibilities are available.

According to the first, the calculations of the characteristics (for example: energy, position) are carried out at the local level, that is to say, in the case of FIG. 7, inside the device 94 associated with the photodetector 60 The results of the calculations are then directed, by a bus common to all the photodetectors to external collection and operating means, for example a microprocessor of a microcomputer. This is the process which has been described above.

According to a second method, the calculations are carried out outside all the architectures cell photodetectors. In this case, all the values stored in the storage means 94 are transferred to an external calculation unit, not shown in the figures. In terms of counting rate, i.e. numbers of events processed per second, the first possibility is advantageous, since only the characteristic quantities associated with the event are to be transferred outwards, at location of all the digital integrals of the photodetectors concerned. In addition, there is then a photodetector calculation system.

The second possibility offers the advantage of not being fixed in the quantities to be calculated, and of being able to modify, for a given quantity, the calculation method. This second possibility, which can be achieved with a programmable system (DSP, microprocessors, dedicated operators, etc.) is more flexible, especially since it is then possible to regain the counting rate by multiplying the transfer means and Calculation.

Particular aspects of the invention make it possible to take account of the possible problems posed by saturated events and / or by stacks.

Saturating events are very high energy events, for example cosmic rays, which can pass through the detector by producing at the output a saturating pulse whose duration can be very long compared to that of the useful signals. In order to avoid these problems, the detected event is invalidated as soon as a sample of the corresponding digitized signal reaches a certain dynamic level, by example the maximum level (255 if coding on 8 bit).

A stack corresponds to the coincidence of two simultaneous or quasi-simultaneous events (separated, in the case of a gamma-camera, for example by less than 1 μs) and close to each other, i.e. - say spaced at most by two rings of photodetectors.

FIG. 9B gives the appearance of the signal resulting from a stack: the waveforms add up.

Consider a simple event of maximum amplitude (Figure 9A). Let t _{3 be} the instant when the signal passes above a threshold S and t ₄ the instant when the signal returns below the threshold. Let ΔT = t ₄ -t ₃ . Now consider any event.

Let t ₃ 'be the instant when the signal passes above the threshold. The event will be considered valid if, at time t ₃ '+ ΔT, the signal is at threshold S or below threshold S. This will be true for a simple event of lesser amplitude; on the other hand in the event of stacking (figure 9B), the signal will pass again at level S only at t ₃ '+ ΔT' and will be above the threshold at t '+ ΔT, since ΔT'> ΔT. In such a case, too, the detected event is invalidated. To avoid all these problems of saturation or stacking, the time when the threshold S is crossed by the signal is identified. As soon as the threshold is crossed, a counter is triggered. If n clock ticks later, n being fixed and corresponding to the duration ΔT, the sample is greater than the threshold, then the event is invalidated: a validity signal (VAL) takes the value "0". ΔT and S are fixed according to the type of event to be detected.

It can be chosen that, if the signal from a photodetector is disabled, it remains there for a certain time, for example 1 μs.

It has been explained above how to take account, for the calculation of a characteristic of an event, of the photodetectors of the first ring (photodetectors 160, 161 in FIG. 3A) around the main photodetector.

It may be important to also take into account the second ring photodetectors.

To this end, as soon as a photodetector is selected as the central photodetector (that is to say that its signal is the largest and that, if a validity test is carried out, it is valid), the means 92 send a "wevt" signal to the local storage system 94 and to the storage systems of the first ring photodetectors indicating that the event must be retained, that is to say that the values of the integrals must be stored and possibly calculate the characteristic quantities of the event. Such a signal is sent synchronously to a clock. As, moreover, an event signal

"evt" is emitted on the previous clock stroke, the resulting time diagram is that of FIG. 10.

After the "wevt" signal has been transmitted to the neighbors of the first ring, these transmit the value of two integrals of second ring photodetectors to the memory 94 of the central photodetector (FIG. 7). Once the values of the digital integrals of the first and second crown photodetectors have been stored, these values can be transmitted to the outside, for an external calculation of quantities of the energy or position type of the event.

Or, in accordance with what has already been explained above, the quantities are calculated by the means 94 themselves and then transmitted to one outside. Both of these possibilities have advantages which have already been indicated above.

The energy calculation will be done for example by applying the formula:

19 E = ∑ Ei i = l

(where i = l, ... 19, designates the 19 values of numerical integrals to be taken into account for a system with a hexagonal structure).

The position will be calculated for example by applying the following formulas, similar to formulas (1) and (2) already presented:

19

Σ ^x iEid

^{X ≈ i} % - ⁽³⁾ ∑Ei i = l

19

∑ YiEi ^

Y = i = l

19 Α)

∑Ei i = l X _x and Yj _. locate the coordinates of the photodetectors of the first and second rings in a reference of the type of that of FIG. 8: Xj _. ef-2, -1, 0.1, 2};

¥, £ {-4, -3, -2, -1,0,1,2,3,4}.

FIG. 11 represents another embodiment of an electronic device 170 associated with a photodetector 60. Such a device is also associated with each of the other photodetectors. A single clock is applied to all the photodetectors and defines the coding step p.

Such a device firstly comprises a first functional unit 188 which performs the following functions, already described above: current-voltage conversion, digitalization of the signal from the photodetector, integration, detection of the passage of the signal by a maximum. In addition, a DMDS signal (equivalent to the EVT signal of the previous embodiment) of storage request is generated at each detection of a maximum and synchronized with the integral of the concerned signal available.

In addition, this functional unit can perform the likelihood-of-pulse test functions by analyzing the amplitude or the duration above a threshold, as explained above.

During each coding step, the device 170 transmits to the corresponding devices neighboring photodetectors 160-1, ..., 160-6 the following data: 1 - the value of the current integral DATA ₀ ,

2 - a validity signal (signal VAL ₀ ) issued when the following two conditions are met: the device 170 is capable of receiving a storage order (storage register 194 available),

* the DATA ₀ signal which it transmits to its neighbors is correct (TRUE tested signal), 3 - a storage decision signal (DS _o signal), the elaboration of which is explained below.

In return, the device 170 therefore receives from each of the neighbors 106-1, .--, 160-6 of the first ring: 1 - the current integral DATAi (i = l, ..., 6),

2 - a validity signal VAL, (i = l, ..., 6),

3 - a storage decision signal DSi (i = l, ..., 6), equivalent to the signal WEVT of the previous embodiment. Thus, at each coding step, the device

170 associated with the photodetector 60 can compare the current value DATA ₀ with that of its neighbors and determine whether it is the largest or not. When an event is detected at a given time (passage of the signal coded by a maximum, greater than a threshold), and when the digital integral of this event is available, the device makes a storage request: a DMDS signal is emitted towards a storage decision generation device 192 (equivalent to the device 92 already described above).

This request is confirmed if, at this time:

1 - the value of the integral of the photodetector 60, which requires storage, is greater than the value of the integral of each neighbor in the first ring,

2 - the validity criterion (signals VAL ₀ and VALi, i = 1-6) is fulfilled for the requesting photodetector as well as for its neighbors in the first ring. When these conditions are met, the storage request is validated (DSo signal). It is then effective in the requesting photodetector and transmitted to its neighbors in the first ring by the intermediary of the signal "storage decision" DS ₀ .

This has the effect of:

1 — in the device 170, of storing in memory 194, the value DATA ₀ of the current integral of the photodetector 60 and the integrals DATA, (i = 1-6) of its neighbors on the first ring,

2 - in each electronic device associated with each of the neighboring photodetectors (first ring) to store the current value of the integrals of two of its own neighboring first ring photodetectors (which are, for some, second ring photodetectors of the requesting photodetector) chosen in such a way that all the current values of the photodetectors of the second ring of the requesting photodetector are retrieved.

When this operation is performed, the system is in the following state:

the storage register 194 of the requesting photodetector 60 has the value of its own digital integral and the value of the integral of the neighboring first ring photodetectors,

- the registers of neighboring first ring photodetectors each have two integral values of second ring photodetectors. The role of the DSi inputs on the device 194 can be specified in the following manner: on the requesting photodetector there can be no DSi active at the time when the photodetector requestor sends a storage decision because the integrals of the first ring photodetectors are smaller than those of the central photodetector, - on the first ring photodetectors, DSi is used to store the value of two second ring integrals.

Thus for hexagonal or round photodetectors arranged in a honeycomb (case of FIG. 3A), the storage register of the requesting photodetector 60 will have stored seven values; each of its six photodetectors on the first ring stores two values, making a total of nineteen values. Similarly for square photodetectors (FIG. 3B), the requesting photodetector will have stored nine values, and each of its eight neighbors two values, ie a total of twenty-five values.

This is true for requesting photodetectors having first and second complete rings. To keep the modularity of the circuit, when the photodetectors do not have all their neighbors (on the edges of the field), the inputs of the missing photodetectors will be fixed at zero.

Means 196 of the sequencer type make it possible to manage the transfer and the calculation and to give the means 198, for generating a busy signal, an end of storage use signal.

The multiplexing means 171 are controlled by the function 196 "transfer management" and provide the following function. When a storage decision has been made, the first ring photodetectors memorized the value of two photodetectors of second crown. To transfer these values to the requesting photodetector, the DATA output bus of each first ring photodetector will be used to reduce the number of links between photodetectors and replace the value of the current integral successively with the two values to be transmitted. The requesting photodetector will thus receive on inputs DATA _a to DATA _f the values of the photodetectors of the second ring. He stores them next to the ones he already had.

The means 192 for generating storage decisions can be implemented as described in FIG. 12.

The means 198 for generating an "occupation" signal can have the form illustrated in FIG. 13, where the reference 193 designates a D-type flip-flop with preset.

The occupation output is put in the logical state (TRUE) by one of the seven storage decision inputs. This state is maintained until the acquired values have been used. The transfer management system then provides a reset pulse.

The functions and system described above can be integrated into an ASIC circuit. In other words, with each photodetector of the network of N photodetectors, there is associated an ASIC circuit.

The digital ASIC of each photodetector may include or provide the following functions: integration, maximum detection, inputs from neighboring photodetectors, output to neighboring photodetectors, comparators, registers of storage, sequencing, device for calculating or a characteristic quantity (s) of an event, reading device.

The digital ASIC can also include the following functions:

- gain multiplication,

- baseline correction,

- management of saturated events.

In order to assess the complexity of the digital ASIC associated with each photodetector, that is to say its number of elementary gates, its surface and its number of input / output pads, we can study its components which require the more doors. These are mainly: • registers,

• adders,

• counters,

• multiplexers,

• multipliers and dividers (for the barycenter). All these components are made up of basic cells that are:

• rocker D,

• the 1-bit adder (FA cell detailed below), • the 2-bit to 1-bit multiplexer.

To which must be added the surrounding combinatorial logic.

The MHS library gives the number of logic gates equivalent to each of these basic cells. So :

• a single clock D flip-flop (DFF) includes 5 doors,

• a 1 bit adder (FA) includes 9 gate, • a 2 to 1 multiplexer (MUX2) includes 3 doors.

An adder of 2 numbers of n bits requires n FA cells ("Full Adder", consisting of 9 gates). From the point of view of the number of doors, a subtractor is to be considered as equivalent to an adder.

To multiply a number of n bits by a number of m bits, with n> m, a multiplier includes n * (ml) FA cells. For example, a 13x13 bit parallel multiplier from the ES2 library has a calculation time of 36.8 ns for an occupied surface of 0.78 mm ² , which is more than enough for the performances envisaged here. The diagram in FIG. 14 gives a schematic overview of the way in which the integration and control entities are arranged (the means 188 of FIG. 12).

The sliding summing element is detailed in FIG. 14, the part ensuring the multiplexing of the output is detailed in FIG. 15.

The summator comprises a shift register with N flip-flops Di, ..., D _N and a summator. The subtractor 220 makes it possible to calculate the difference between the input and the output of the shift register.

The adder 221 accumulates these deviations in the register 223. A base line detection circuit 222 composed of a comparator and a shift register controls a storage register 225 of the base line. This is then subtracted by the subtractor 226 from the current value of the sliding sum. The means 224 which ensure the detection of events and the concept of validity are essentially composed of a maximum detection, carried out using a comparator, followed by a shift register necessary to synchronize the event signal with the output of the corresponding integral. The sliding summator is composed of a shift register (10 8-bit registers or 10 * 8 = 80 D flip-flops), a 12-bit register (12 D flip-flops), a 12-bit summator and an 8-bit subtractor, i.e. (12 + 8) = 20 FA cells. The summator therefore includes 92 DFF and 20 FA.

- Determining the baseline requires comparison logic and a 16-bit shift register with reset, or 16 DFF.

Add the 12-bit offset register and the 12-bit subtractor, which corresponds to 12 DFF and 12 FA.

The multiplexing part of the output can be shown diagrammatically as illustrated in FIG. 15.

It has 2 registers of 12 bits, (24 DFF), 1 multiplexer with 6 inputs of 12 bits, and 1 multiplexer with 8 inputs of 12 bits.

However, to make a MUX with 6 inputs, we need 5 MUX with 2 inputs. Similarly, it takes 7 MUX with 2 inputs to make an MUX with 8 inputs. So in total you need (7 + 8) * 12 = 180 MUX2.

In summary, the integration and multiplexing entity includes:

• 144 DFF, or 144 * 5 = 720 doors,

• 32 FA, or 32 * 9 = 288 doors, • 180 MUX2, or 180 * 3 = 540 doors.

This brings the number of doors to a total of around 1600 doors, including the surrounding combinational logic. The control entity (92, 192 in FIGS. 7 and 12) comprises, in particular:

• 4 8-bit comparators (approximately 4 * 8 * 5 = 160 doors).

• A 10-bit counter and a 6-bit counter, ie 16 DFF, therefore 80 doors,

• if we add all the sequencing logic, we can evaluate the command entity at 300 doors.

The comparison entity (190-1, ..., 190-6 in FIG. 11) comprises:

• 6 12-bit comparators, or 360 doors,

• 19 12-bit registers for energy storage, or 228 FFD including 1200 doors.

With regard to the calculation entity of the barycenter, we have already seen that this calculation can be done by applying the formulas (1), (2) and (3), (4) above. The calculation structure therefore includes:

• a 12 * 4 bit multiplier, or 36 FA,

• a 14-bit register, ie 14 DFF,

• a 14-bit summator, or 14 Fa,

• 3 other 14-bit summers are needed to obtain the sum of the 18 values, ie 42 FA,

• a result storage register on 14 bits (X 'or Y'), ie 14 DFF.

This structure is to be repeated twice since there are two coordinates, so in total we have: • 384 FA, or 3,460 doors,

• 112 DFF, ie 560 doors, therefore 4000 doors. We therefore need two parallel multipliers, a 14 * 13 bit and a 14 * 6 bit, which represents 238 FA.

You need a 27-bit register and a 20-bit register to store the results, or 47 DFF. This gives a total of 2,400 doors.

It remains to divide these numbers by the total energy, to obtain the final coordinates. For that, one can be satisfied to carry out a sequential division in 9 times of clock. To this end, we can evaluate that, to divide 27 bits by 14 bits and obtain 9 bits, we need:

• 1 27-bit register with 27 MUX2,

• 1 14-bit register, • 1 9-bit register,

• 1 14-bit subtractor.

This is equivalent to 50 DFF, 14 FA and 27 MUX2, making a total of around 500 doors.

You need two dividers in total, which makes 1000 doors.

The "transfer" entity 196 (FIG. 11) requires three registers to store X, Y and E, that is to say two registers of 9 bits and 1 of 14 bits, which is equivalent to 32 DFF, that is to say 160 gates. If we add the sequencing logic, we can evaluate the entity at 200 doors.

The ASIC is thus evaluated at 11,000 doors. In 0.7 μm technology, the number of doors per mm ² is around 1400 doors / mm ² . We can therefore assess the area occupied by the doors at 8 mm2, or approximately

3 mmx3 mm.

A 140-legged housing can be used. The invention described applies to a gamma camera as described in the introduction to the request, in conjunction with FIG. 1. It advantageously applies to the implementation of the techniques of "correction of attenuation by transmission "and" PET (Position Emission Tomography) coincidentally, already described in the introduction to the present application.

The invention has been described using the digital integral as a signal representative of the intensity of each signal coming from each photodetector.

In fact, it is also possible to use, instead of the digital integral, the maximum intensity of the digital signal or of the analog signal. However, the digital integral is the quantity which is best suited to the implementation of the invention.

If we want to implement a device in which the maximum intensity of the digital signal or of the analog signal is chosen as a quantity representative of the intensity of each signal coming from each photodetector, it suffices to replace, in the chain of detection associated with each photodetector, the means for calculating a digital integral by means for detecting a digital or analog maximum. The management of the data obtained (comparison, storage) is then done in the same manner as described above.

Claims

1. Method for processing signals produced by a set of N photodetectors in response to an event to be detected, comprising: - the comparison, for each photodetector, of a signal representative of the intensity of the digitized signal obtained from the transmitted signal , with a threshold value,

the comparison, for each photodetector emitting a signal whose intensity exceeds said threshold, of the intensity of the signal that it emits with the intensity of the signal emitted by each of the neighboring photodetectors, called first ring,

- memorizing the value of the maximum intensity among all the intensities compared, as well as the intensities of the signals emitted by the photodetectors close to the photodetector whose signal has the maximum intensity.

2. Method according to claim 1, the signal representative of the intensity of the digitized signal being either a digital integral of the digitized signal, or the maximum intensity of the digitized signal, or the maximum intensity of the analog signal.

3. Method according to claim 1 or 2, at least one signal, representative of a quantity characteristic of the event, being produced using the stored intensities.

4. Method according to claim 3, a signal representative of the energy of the event being produced.

5. Method according to claim 3, a signal representative of the position of the event with respect to all of the N photodetectors being produced.

6. Method according to claim 3, at least two signals representative of the barycentric coordinates of the event with respect to the set of N photodetectors being produced.

7. Method according to one of claims 1 to

6, an event signal being emitted, for each photodetector, when the signal representative of the intensity of the digital signal is formed.

8. The method of claim 7, the intensity of the digital signal being represented by its digital integral, the latter being calculated over a certain number of clock steps, from the maximum of the digital signal, and the event signal being emitted a certain number of clock steps after passing through the maximum.

9. Method according to one of claims 1 to

8, further comprising a step of detecting and rejecting saturating events and / or stacks of events.

10. Method according to one of claims 1 to

9, further comprising a step of calculating and storing the intensities of the so-called second ring photodetectors.

11. Method according to one of claims 3 to 6 and according to claim 10, the signal representative of a quantity characteristic of the event being produced using the stored intensities of the first and second crown photodetectors.

12. Method for identifying a characteristic of an event occurring before a set of N photodetectors, comprising:

- the identification of a subset of Ni photodetectors, concerned by the interaction, ie ie having produced a signal in response to the event, - the production of a signal representative of the characteristic of the event, as a function of the intensities:

* of the maximum intensity signal among the Ni signals produced by the Ni photodetectors, in response to the event,

* and signals emitted by the neighboring photodetectors (called the first ring) of the photodetector whose signal has the maximum intensity among the Ni photodetectors.

13. The method of claim 12, the characteristic being the energy of the event.

14. The method of claim 12, the characteristic being the position of the event with respect to the set of N photodetectors.

15. The method of claim 14, the position being identified by a pair of barycentric coordinates.

16. The method of claim 12, the signal representative of the characteristic of an event being produced as a function, in addition, of the intensities of the photodetectors of the second ring.

17. Method according to one of claims 12 to

16, the intensity of the signal produced by each photodetector being either a digital integral of the digitized signal, or the maximum intensity of the digitized signal, or the maximum intensity of the analog signal.

18. Device for processing signals produced by a set of N photodetectors, comprising: means for comparing, for each photodetector, the intensity of a signal transmitted with a threshold value,

means for comparing, for each photodetector emitting a signal whose intensity exceeds said threshold, the intensity of the signal that it emits with the intensity of a signal emitted by each of the neighboring photodetectors,

- Means for memorizing the value of the maximum intensity among all the intensities, as well as the intensities of the signals emitted by the photodetectors close to the photodetector whose signal has the maximum intensity.

19. Device according to claim 18, further comprising means for producing a signal representative of a quantity characteristic of an event using the intensities.

20. Device according to claim 19, at least one signal representative of the energy of the event, or of the position or barycentric coordinates of the event with respect to the set of N photodetectors, being produced.

21. Device according to one of claims 18 to 20, further comprising means for transmitting an event signal, for each photodetector, when the intensity is calculated.

22. Device according to claim 21, comprising means for detecting the passage of the signal by its maximum and for emitting the event signal a certain time after the passage by the maximum.

23. Device according to one of claims 18 to 22, further comprising means for detecting and rejecting saturated events and / or stacks of events.

24. Device according to one of claims

18 to 23, comprising means for calculating and storing the intensities of the second ring photodetectors.

25. Device according to one of claims

19 or 20 and according to claim 24, the signal representative of a quantity characteristic of the event being produced using the intensities of the photodetectors of the first and second rings.

26. Device according to one of claims 18 to 24, comprising either means for calculating a digital integral of the signal emitted by each photodetector, or means for identifying the maximum intensity of a digital signal corresponding to the signal emitted by each photodetector, or to identify the maximum intensity of the analog signal emitted by each photodetector.

27. Device for identifying a quantity characteristic of an event occurring in front of a set of photodetectors, comprising:

- means for identifying a subset of Ni photodetectors, having produced a signal in response to the event, - means for producing a signal representative of the quantity characteristic of the event, as a function of the intensities:

* the maximum intensity signal among the Ni signals, produced by the Ni photodetectors in response to the event,

* signals emitted by neighboring photodetectors, called first ring photodetectors, of which the signal has the intensity among the Ni photodetectors.

28. Device according to claim 27, the characteristic quantity being the energy of one event.

29. Device according to claim 27, the characteristic quantity being the position of the event with respect to the set of N photodetectors.

30. Device according to claim 28, the position being identified by a pair of barycentric coordinates.

31. Device according to one of claims 27 to 30, the intensity of each signal produced by each photodetector being either a digital integral of each signal, or the maximum intensity of the corresponding digital signal, or the maximum intensity of the analog signal.

32. Device according to one of claims 19 or 27, the means for producing a signal representative of a characteristic quantity being specific to each photodetector.

33. Device according to one of claims 19 or 27, the means for producing a signal representative of a characteristic quantity being common to all the photodetectors.

34. Device for the detection of events comprising:

- a two-dimensional network of N photodetectors, - a device according to one of claims 18 to 33.

35. Camera comprising a device for detecting events according to claim 34, the photodetectors being photomultipliers.

36. A gamma radiation imaging device, comprising a camera according to claim 35.

37. Device according to one of claims 18 to 33, the functions of detecting the intensity of the signal emitted by each photodetector, of comparing with neighboring photodetectors, of storing and / or producing a signal representative of a event being ensured by an ASIC associated with each photodetector.