ITTO20130128A1 - PHOTON SENSOR DEVICE - Google Patents

Description

DESCRIPTION of the industrial invention entitled: "Photonic sensor device"

DESCRIPTION

TECHNICAL FIELD

The present invention relates to a photonic sensor which is, for example, suitable for asynchronous detection and measurement of scattered photonic events. The sensor in some embodiments finds application whenever short (for example, a few hundred ns) light pulses are generated and it is desired to accurately capture the spatial position and time of arrival, as well as determine the number of photons in the event. An application? that in the field of Positron Emission Tomography. However the invention is not? limited to the detection of gamma rays and can? also be applied to many areas of biophotonics, quantum photonics, X-ray photonics, spectroscopy and particle physics.

STATE OF THE ART

Positron Emission Tomography (PET)? a nuclear imaging technique that provides functional three-dimensional images of the body, and finds its main applications in clinical oncology and brain function analyzes. PET imaging systems involve the detection of coincident and diametrically opposite pairs of gamma photons, which are generated by the annihilation of a positron emitted in the decay process of a radioactive tracer injected into a patient or other subject. These gamma photons are absorbed by scintillator crystals, which emit a faint pulse of visible photons for each absorbed gamma photon.

An efficient PET system must measure the spatial position of the event, the precise arrival time of the scintillation light pulse (for a coincidence recognition), and the number of photons detected, allowing the evaluation of the energy of the incident gamma photon. (in order to be able to discard detections with incorrect energy, such as background radiation or scattered rays).

Not only does the detector in a PET imaging system need to be sufficiently sensitive, since it does not need to be sensitive enough. the number of photons emitted? extremely low, but it must also provide time domain information about the arrival of incident photons to meet the above requirements. In a traditional system, each scintillator 2? coupled to a photon multiplier tube ("Photon Multiplier Tube"? PMT) 4, forming a detector module, with many of these modules forming a ring 6 around the patient 8, how? shown schematically in Figure 1.

While PMTs are a developed technology, which exhibits excellent sensitivity, they are nevertheless subject to a number of important disadvantages in that they are incompatible with magnetic fields, expensive, bulky, fragile, and require very high polarization voltages. . The first of these disadvantages, the incompatibility? with magnetic fields, constitutes a significant barrier to the development of combined PET scanners? MRI, which would allow the simultaneous execution in a single device of the generation of both physiological and functional images, without the possibility? of patient movement or alignment problems that blur the resulting image, a long-standing goal of the community? which deals with the generation of medical images. The cost of PMT detectors in a typical system represents a significant fraction of the overall cost of the equipment, which limits the number of institutions that can afford the equipment. Finally, the remaining limitations increase the footprint of the complete equipment.

In recent years, yes? There is a growing interest in building a robust, low cost, mass-producible CMOS alternative to the PMT, the long-standing detector of choice in many nuclear imaging applications, such as PET, in order to eliminate the limitations discussed previously. Silicon photomultipliers ("Silicon Photon-Multiplier"? SiPM) provide a direct alternative to the PMT, comprising a Single Photon Avalanche Diode array ("Single Photon Avalanche Diode"? SPAD), also called Avalanche Photo Diodes (" Avalanche PhotoDiode "? APD) in mode? Geiger, electrically coupled to provide analog output.

A CMOS approach can? potentially provide several advantages over PMTs. For example, CMOS-based systems can be compatible with magnetic fields allowing modality? do you combine PET? MRI. Such systems can also be more? cheap (for example, manufactured using a CMOS process in large series) and more? small (for example, the thickness of the sensor may be less than 1mm, compared to ~ 100mm for many PMTs). Also, an analog and digital signal processing on the "chip" can? eliminate the need? of various external components, resulting in a reduction in power consumption, system size and bill of materials.

Bench? there are many solid state SiPM applied to PET and produced in processes associated with CMOS, these sensors often do not contain a capacitance. of significant processing built-in and are normally unable to provide a map of photon arrival times and positions.

US 7,723,694 discloses an example of a photonic sensor device using photodiodes, in this case used in a time-of-flight positron emission tomography scanner. A plurality? of detectors are arranged in rings around a patient. Each detector comprises a scintillator and an associated matrix of detector pixels, with each detector pixel providing, through a pixel level reading circuit, a digital representation of a photon count and a time-stamp. ) for each detected scintillation event. Each pixel comprises a matrix of detector cells, each of which includes a photodiode. Scintillation detection events are triggered in response to the crossing of a threshold by a signal from one of the detector cells. However, by activating outputs from individual detector cells, the device? sensitive to defective detector cells, and to false positive signals. US 7,723,694 describes the inhibition of defective cells, but this can? be time-consuming and can in turn be prone to errors. Additionally, cell defects can develop over time, requiring repeated defect monitoring. In addition, the equipment provides intensity data? and position for each triggered event, but does not allow monitoring of time profiles of events or of the periods preceding the events.

Other examples of known photonic sensor devices based on diodes are described in US 7,626,389 and in WP 2006 / 111.883.

SYNTHESIS

In a first aspect of the invention, a sensor device is provided for the detection of photons, comprising a plurality of photons. of pixels, with each pixel comprising a plurality of pixels of photon detectors and a processing logic able to output a respective detection signal for this pixel for each interval of a series of time intervals ("time bin" in the Anglo-Saxon technical literature), a combiner suitable, for each time interval , to combine the detection signals from each pixel to produce a combined detection signal for this time interval, and an event monitoring device for identifying the occurrence of a detection event during a particular time interval as a function of the value of at least one of the combined detection signals. The determination can? be carried out as a function of the value of the combined detection signal for this time interval or, for example, as a function of the values for this time interval and / or for at least one other preceding or following time interval.

The sensor device can? operate in accordance with a first modality? before identifying a detection event, and? able to enter a second mode? in response to the event monitoring device having identified the occurrence of a detection event. The first mode? can be indicated as a mode? inactive. The second mode? can be a mode? event monitoring.

The device can? further comprising means for obtaining an integrated detection signal for each pixel, representative of accumulated detection signals during a second mode integration period.

The processing logic for each pixel can? include a respective integrator, and, in operation, the integrator for a pixel can? get the built-in detection signal for this pixel.

The sensor device can? be able to pass from the second mode? to the first mode? when ? at the end of the integration period.

For each pixel, the processing logic can? include at least one time stamp device which? adapted to provide a time stamp to record the detection time of one or more? photons by the detectors of this pixel.

The time stamp can? record the detection time with a resolution pi? end of the length of the time interval.

The device can? further comprising means for determining an event time from the time stamp data obtained from each pixel of the plurality. of pixels, for example by calculating the average of the time stamp data obtained from the plurality? of pixels.

Said at least one time stamping device of a pixel can? be adapted to provide a single time stamp for each time interval for which at least one photon is detected.

For each pixel, the processing logic can? comprise a first time stamp device and a second time stamp device, and the first and second time stamp devices can be used alternately, so that the first time stamp device is used to provide a time stamp during some time slots, and the second time stamp device is used to provide a time stamp during other time slots.

The device can? furthermore comprising, for each pixel, a data storage medium which? adapted to store at least one detection signal and / or at least one time stamp for at least a previous time interval for this pixel.

For each pixel, the data storage medium can? be adapted to store, upon identification of the occurrence of a detection event, at least one detection signal and / or at least one time stamp which precedes, during, or which follows the identification of the occurrence of the detection event.

This characteristic ? particularly important and therefore, according to a further independent aspect of the invention, a sensor device for the detection of photons is realized, it comprises a plurality of of pixels, with each pixel comprising a plurality of pixels of photonic detectors and a processing logic adapted to output a respective detection signal for this pixel for each interval of a series of time intervals, and an event monitoring device for identifying the occurrence of a detection event during an interval particular time, in which, for each pixel,? present a data storage medium that? adapted to store at least one detection signal and / or at least one time stamp for at least a time interval which precedes, during, or which follows the identification of the occurrence of the detection event.

The event monitoring device can? be able to include in the integrated detection signal of the second mode, for example the mode? event monitoring, the detection signals stored for at least one time interval preceding, during, or following the identification of the occurrence of the detection event.

For each pixel, the data storage medium can? comprising at least one FIFO for storing said at least one detection signal and / or at least one FIFO for storing said at least one time stamp.

The event monitoring device can? be able to send one or more? control signals ("trigger signals") to the processing logic of each pixel in response to the identification of a detection event, to block the operation of the at least one FIFO of this pixel while in the mode. event monitoring.

For each pixel, the processing logic can? be able to respond to one or more? control signals ("trigger signals") for controlling the integrator of this pixel to accumulate the integrated detection signal for the integration period.

The combiner can? comprise a tree structure comprising a plurality of of nodes in a plurality? of levels, in which the pixel processing logic provides pixel detection signals to the root nodes of the tree and the tree structure? realized in such a way that, in each level node above the root node, the signals of at least two nodes at a lower level of the tree structure are combined.

This characteristic ? particularly important and therefore, according to a further independent aspect of the invention, a sensor device for detecting photons is provided, comprising a plurality of photons. of pixels, with each pixel comprising a plurality of pixels of photonic detectors and a processing logic adapted to output a respective detection signal for this pixel for each interval of a series of time intervals, a combiner able, for each time interval, to combine the detection signals of each pixel in so as to produce a combined detection signal, wherein the combiner comprises a tree structure comprising a plurality of of nodes in a plurality? of levels, in which the pixel processing logic provides pixel detection signals to the root nodes of the tree and the tree structure? realized in such a way that, in each level node above the root node, the signals of at least two nodes at a lower level of the tree structure are combined.

The combination of signals can? understand the sum of signals, and each node can? understand an adder.

The nodes can operate in accordance with a periodic timing of the sensor, and the nodes can be arranged so that, in each node, the signals that are combined are signals that correspond to the same time interval.

The nodes can be arranged so that, for all nodes at each level of the tree structure, the times for the arrival of signals from nodes of the previous level of the tree are substantially the same. What,? equal routing delay can be achieved.

The nodes can be arranged so that, for all nodes at each level of the tree, the distances from the nodes of the previous level of the tree are substantially the same.

The tree structure can? comprising an output node which outputs, for each time interval in succession, a combined detection signal, wherein the combined detection signal for a time interval represents a total of the signals obtained from each of the pixels for this time interval.

The combiner can? understand a hierarchical tree arrangement of adders.

The device can? further comprising means for obtaining, for each detection event, a profile of the combined detection signal as a function of time.

The device can? further comprising means for determining an instant when the detection event occurs by analyzing the profile of the combined detection signal as a function of time.

The event monitoring device can? be able to obtain, for each detection event, one or more, possibly each, of the following parameters: a profile of the combined detection signal for all the pixels as a function of time during the integration period; an integrated detection signal for each pixel, representative of the detection signals accumulated by this pixel during the integration period; time stamp data for each pixel.

For each pixel, the processing logic can? include at least one time stamp device which? adapted to provide a time stamp to record the instant of detection of one or more? photons by the detectors of this pixel.

The time stamp can? be included in the detection signal for this pixel and for a particular detection time period.

The time stamp can? recording the instant of detection of the first photon during a detection time interval, or the detection instant of the second photon, of the third photon or the detection instant of a photon occupying any ordinal position during a particular time interval.

The time stamp can? record the instant of detection with a resolution pi? end of the length of the time interval.

For each pixel, the processing logic can? comprise a first time stamp device and a second time stamp device, and the first and second time stamp devices can be used alternately, so that the first time stamp device is used to provide a time stamp during some time slots, and the second time stamp device is used to provide a time stamp during other time slots.

For each pixel, the detection signal can? comprising a count signal representative of a number of photon detection events.

For each pixel, the processing logic can? be able to combine signals from the plurality? of detectors to obtain the pixel detection signal. For example, the processing logic can? be able to sum the signals provided by the plurality? of detectors to obtain the pixel detection signal.

Each detector can? comprise a single photonic detector and / or photodiode detector, optionally a Single Photon Avalanche Diode (SPAD).

Each pixel can? understand a plurality? of Silicon Photo-Multipliers (Silicon Photon-Multiplier "? SiPM), each of which includes a plurality of detectors.

The device can? be included in a chip, such as a CMOS chip.

The device can? furthermore comprising a processing resource included on the chip and able to determine at least one parameter between:

the instant in which the detection event occurs;

a location of the detection event;

a pixel count profile as a function of position for the detection event;

a sum of all integrated detection signals (spatial data counts) from one or more? pixels, to provide, for example, an energy estimate for the detection event.

The detection event can? include at least one event between a scintillation event, a particle annihilation event, a fluorescence event, or the reception of photons arising from such a scintillation event, particle annihilation event, or fluorescence event.

According to a further independent aspect of the invention, a detection system is provided comprising a plurality of products. of sensing devices as claimed or described herein, each of which? aligned with a scintillator material, and a processor for processing detection signals and / or integrated detection signals obtained from the sensor devices.

The system can? be a PET, MRI, X-ray, particle physics, quantum photonics, or biophotonics detection system.

According to a further independent aspect of the invention, a photonic detection process is carried out, comprising the acquisition of a respective detection signal from each pixel forming part of a plurality of materials. of pixels for each interval forming part of a series of time intervals, in which each pixel comprises a plurality of of photonic detectors, for each time slot, combining the detection signals from each pixel to produce a combined detection signal for this time slot, and identifying the occurrence of a detection event during a particular time slot as a function of a value of the combined detection signal at least for this time interval.

According to another independent aspect of the invention, a photonic detection process is carried out, comprising the acquisition of a respective detection signal for each pixel forming part of a plurality of products. of pixels for each interval forming part of a series of time intervals, in which each pixel comprises a plurality of pixels. of photonic detectors; for each pixel, storing at least one detection signal and / or at least one time stamp from at least one previous time interval and identifying the occurrence of a detection event during a particular time interval, where, for each pixel, the at least one stored detection signal and / or the at least one time stamp comprises at least one detection signal and / or at least one time stamp for at least one time interval preceding, during, or following the identification of the occurrence of the detection event.

According to a further independent aspect of the invention, a sensor is provided for asynchronous detection of a photon pulse composed of a pixel matrix of silicon mini-photomultipliers ("mini-Silicon PhotoMultiplier" - mini SiPM). Each pixel contains a matrix of single photon detectors arranged in the form of a silicon photomultiplier ("Silicon PhotoMultiplier" - SiPM) and a processing logic synchronized by a global clock waveform. The logic generates two outputs: (1) a digital photon count that counts all the photons incident on the pixel surface within the clock period, (2) a digital 'time stamp' representing the instant of arrival of the first photon with respect to the present clock period. Various pixel detector arrangements (such as that of the FBK patent), dSiPM or multi-pixel photon counting arrangements ("Multi-Pixel Photon Counting"? MPPC) are possible with a counting logic that provides a single digital photon count. .

Each pixel of the further embodiment can? to include a first-in buffer? first out ("First-In First-Out"? FIFO) in which digital photon counts and time stamps are entered and translated at each global clock cycle. The FIFO? a fundamental component of the pixel since? allows you to keep a local history of the energy and timing of the photons detected until? a global mechanism invalidates an event. The FIFO of all the pixels can? be locked (frozen) at the assertion of a global FREEZE signal, so that no further inputs are provided to the FIFO until? the FREEZE signal is not lowered.

Each pixel can? including an integrator which sums the digital photon count output for a programmable time interval from the FREEZE signal assertion. Integrated photon counting? available as a third digital output from the mini-SiPM.

Each pixel can? include one or more? time stamp generating circuits (for example digital time converters) which can assign a time stamp to the first or second or third photon and so on? away, or any combination of them.

Pixel outputs can be summed hierarchically under the control of a global clock. The outputs of pairs of neighboring 'cells' can be added during each clock period and the sum can be added. constitute the exit to the next level of hierarchy. The 'cell' pi? simple ? the pixel that outputs the total number of photons incident on its surface (typically a small multiple of 100? m? 100? m). The hierarchical sum of the pixel outputs creates a single higher-level digital output from the sensor that represents the total number of incident photons within a sampling period over the entire surface of the sensor (e.g. 10mm? 10mm). The number of stages of hierarchical sum? defined by log2 (number of pixels in a sensor). The network of hierarchical sums can? be implemented in the form of an H-tree spatial network to maintain equal routing delays between all cells. The summing operations can preferably be carried out within the sensor area so as to maintain the maximum overall photon sensitive area. Other global sum approaches are also possible that aggregate the output of all pixels. The distribution of the global clock can? also assume the same H-tree arrangement in the opposite direction (from sensor input to pixel) to minimize temporal misalignments between pixels across the matrix.

The higher level digital output can? be compared to a threshold (for example, a digital input word or 'trigger threshold') in each clock cycle to produce a single-bit trigger sequence, for example asserted if the output of the sensor is above the activation threshold and otherwise lowered.

When the activation threshold is crossed and a FREEZE signal is asserted and fed back to the pixel matrix, ci? can causing the pixel FIFOs to be inhibited together with the integration of the pixel photon counts for a variable number of clock periods after the FREEZE signal assertion. In a PET / MRI application, sensor activation can? depend on the arrival of a gamma photon with sufficient energy that impinges on the scintillator. The cloud of photons emitted in the visible can? be counted by the sensor and compared with a threshold to allow to distinguish the dark count of the sensor from a short photon pulse (typically 160 ns duration) generated by a gamma scintillation in a scintillator crystal placed above the sensor. The FIFO of the pixels can allow there to be a certain number of clock cycles of latency between the arrival of the gamma photon and the generation of the activation signal (which requires the sum of all the photons arriving on the sensor) without loss of markings. thunderstorms or counts that occur immediately after the gamma event.

Digital time stamps and integrated photon counts (referred to for example as 'pixel data') can be read from each pixel. There? can be obtained by generating a pixel enable signal as a result of which the pixel data is positioned on an output bus of the sensor. The FREEZE signal of pixels can? be lowered when data from all pixels in the sensor has been read.

Any feature in one aspect of the invention can? be applied to other aspects of the invention, in any appropriate combination. For example, device characteristics can be applied to process characteristics and vice versa.

BRIEF DESCRIPTION OF THE DRAWINGS

Some embodiments will now be described, by way of non-limiting example only, and are illustrated in the following figures, in which:

Figure 1 shows a schematic diagram of a known PET image generation system;

figure 2 shows a schematic diagram showing in an overall view the architecture of a sensor chip according to an embodiment;

Figure 3 shows a schematic illustration of a pixel in the embodiment illustrated in Figures 1 and 2;

Figure 4 shows a schematic diagram showing a processing logic of a MiniSiPM included in the pixel illustrated in Figure 3;

figure 5 shows a schematic diagram which illustrates more? in detail a processing logic of the pixel illustrated in Figure 3;

Figure 6 shows a time diagram, and a histogram of total photon counts obtained before and during a detection event;

figure 7 shows a time diagram pi? detailed;

Figures 8a and 8b show alternative arrangements of the pixel logic;

Figures 9 to 11 show illustrations of semiconductor structures forming a processing logic according to some embodiments;

Figure 12 shows a normalized graph of the energy as a function of occurrences obtained from measurements using the embodiment illustrated in Figure 2, and comprises an interpolated peak; And

Figure 13 shows a graph of the number of occurrences as a function of time obtained from TDC codes generated by a single pixel of the embodiment illustrated in Figure 2, compared with a reference system and with an average of frames.

DETAILED DESCRIPTION

The higher level architecture of a sensor device, in the form of a sensor chip 10, according to an embodiment? shown schematically in Figure 2. In the use of a PET sensor, the chip 10? positioned with respect to a scintillator crystal (not shown) so that gamma photons generated by the annihilation of a positron in the scintillator crystal are detected by the sensor chip 10. A number of scintillator crystals and associated detector chips 10 can be detected. be arranged in a cylindrical configuration similar to that illustrated in Figure 1.

Chip 10 constitutes a 92k SPAD sensor, 16? 8 pixels, in a low voltage CIS CMOS technology at 0.13µm, with each pixel comprising four mini-SiPMs of 180 SPAD each, in-pixel counters, and Time Converters? Digital ("Time to Digital Converter"? TDC). This sensor architecture can? provide a capacity? real-time energy histogram of 100M samples / s and an on-chip discriminator for gamma event recognition. The structure of the chip? described more? in detail with reference to Figures 2 to 5.

The sensor chip 10 comprises an array of pixels 20 which are connected to an event monitoring device which includes a logic circuit in the form of a discriminator 22. The discriminator 22? also connected to configuration registers 24 which store the values of various configuration parameters, and to a row and column digital address decoder 26 which allows the reading of detection signals from single pixels 20, if desired. ? also provided for a SRAM 28 decoder that can? be used to selectively enable / disable each SPAD, if desired. The device also comprises a sensor output bus 30 which includes serializer and MUX components and which permits transmission of sense data and other data to a central processor (not shown).

Discriminator 22 comprises a clock (not shown) which provides a CLKBIN clock signal, with an operating frequency of 100 MHz. The sensor chip? designed to work in synchronism with the CLKBIN clock signal.

Each pixel 20 comprises a processing logic, how? discussed more? in detail below with reference to Figures 3 to 5. The pixels 20 are connected to the discriminator 22 through a tree structure, in the form of a hierarchical H-tree 32 which? superimposed on the pixel matrix.

The H-tree structure ensures that, from the periphery, the main CLKBIN clock and the critical signals? time are provided to the pixels 20 with equalized propagation delays. In the opposite direction, an adder? disposed in each node of the H-tree structure and all the pixel detection signals, in this case in the form of pixel counts, in each clock period are summed by the resulting distributed adder. What? a combined detection signal, in this case a chip-level photon count,? generated for each time interval of the clock period. A time interval can? be referred to as a detection time interval.

A feature of the embodiment illustrated in FIG. 2 is that the discriminator 22 controls the operation of the chip so that it has one of two modes, referred to in this case as a mode. inactive and a mode? event monitoring, depending on whether a detection event (in this case, a gamma event) has occurred. The use of two modes? provides the required temporal, spatial and energy information relating to each gamma photon detected, while at the same time avoiding the generation of such information by SPAD dark counts. A further feature of the embodiment is that the discriminator 22 determines whether yes? or not? a detection event occurred for a particular detection time interval (time interval) as a function of the combined detection signal value on all pixels for this detection time interval.

While waiting for a gamma event in the mode? inactive, all the photons detected by each pixel are counted during a period of 10 ns (defined by the clock distributed in the H-tree, called CLKBIN). These pixel counts are summed hierarchically to produce a single 16-bit total representing the instantaneous photon flux affecting the sensor pixels, with a temporal resolution of 10 ns. This real-time histogram data is made available outside the chip through a parallel 16-bit output bus, while a discriminator on chip 22 applies logic that compares photon counts for two consecutive intervals with two configurable thresholds (Th1 , Th2) to distinguish a gamma event (which can, for example, produce about 1,000 photons in a 200 ns window) from noise (dark and spurious photon counts), as schematically illustrated in the discrimination process flowchart provided in Figure 2.

If the total count (Tot cnt)? higher than the appropriate threshold (Th2), then the discriminator 22 considers the light incident at the moment on the sensor as a valid gamma scintillation event and the device switches from the mode? inactive to the mode? event monitoring. The detection signals are then integrated until? the time (Tmr) not? equal to the integration period (IntTime), as described more? come on.

? Any other appropriate threshold application procedure can be used. Using a threshold application procedure or other procedure based on the combined detection signal obtained from all pixels, the discriminator 22? able to reduce the impact of false signals, or malfunction, of individual pixels or individual photonic detectors, in determining when? a detection event occurred.

The differences in the operation between the mode? inactive, when waiting for a detection event, and the mode? event monitoring, after which? an event has been detected, are described more? in detail below. First, the pixel structure and associated processing logic are further described.

A pixel 20? shown schematically in Figure 3. Each pixel 20? composed of four independent digital SiPMs (MiniSiPMs) 40a, 40b, 40c, 40d, and of a processing logic comprising a common data management circuit ("DAta MAnaging Circuit" - DAMAC) 42 responsible for photon counting, time stamping and of data transmission.

Each MiniSiPM 40a, 40b, 40c, 40d comprises a honeycomb matrix of 12? 15 SPAD 44 individually digitized together with the respective processing logic 46a, 46b, 46c, 46d and a compression circuit. The SPAD 44 in this embodiment have an active diameter of 16.27µm.

The processing logic 46a of one of the MiniSiPM 40a? represented schematically in Figure 4, and includes a 'front-end' 50 for each SPAD, a spatial-temporal compression circuit 52, and counters 54. Within the MiniSiPM 40a, all the cathodes of the SPADs are connected to a common bias line , while their anodes are individually connected to a passive inhibition transistor ("quenching" in the Anglo-Saxon technical literature) M1. The release of the SPAD? digitized by a Schmitt triggered inverter to prevent the slow charging of the SPAD 44 from affecting the compression circuit, while the SRAM 6Ts are used to inhibit noisy SPADs 44.

Finally, the compression circuit 52 combines the outputs of the 180 SPADs into a single channel (Vpulse) which provides a short pulse for each photon. The relatively long pulses of the SPADs (tens of ns) are first subjected to a logical OR operation in groups of three (spatial compression) and then compressed over time by the monostable.

The speed? maximum of the Vpulse output channel? defined by the pulse width of the monostable, which can? be reduced to 250 ps in this implementation. Vpulse? then fed to the DAMAC 42, which constitutes a part of the pixel processing logic, by means of a time interval sampling circuit.

The DAMAC 42, which is part of the pixel processing logic,? illustrated schematically pi? in detail in figure 5. Does the DAMAC 42 include two time converters? 12-bit digital ("Time to Digital Converter"? TDC) 60a, 60b, two counters 62a, 62b, a hierarchical adder 64 operating to sum the photon counts received for each of the four SiPM 40a, 40b, 40c, 40d , a timestamp FIFO 66, another FIFO 68 which also functions, in part, as an accumulator during the mode? event monitoring, and an output to a communication bus. The output of the hierarchical adder 64? also connected to a hierarchical summing tree of the chip which connects the outputs of each of the pixels 20 of the chip 10, how? discussed more? in detail below.

Each DAMAC 42? adapted to receive the main clock signal CLKBIN, which operates as a global clock signal for each of the pixels 20.

In operation, each period of the main clock CLKBIN? assigned to time intervals A or B, selecting one of the two counters 62a, 62b and of the two TDCs 60a, 60b, operating in a mode? alternating operation. During each time interval, while the pulses of the MiniSiPMs are accumulated in a first counter (e.g. counter A) 62a, the output of the other counter (counter B)? available for the hierarchical adder in the pixel (adder 1, 2 and 3), which calculates the total number of photons in the pixel. Similarly, one of the two TDCs (e.g. TDC A, 60a) measures the arrival time of the first photon in the pixel in the current interval, while the other (TDC B, 60b) blocks the sampled value of the previous interval . The outputs of both the adder and the TDC of the previous time interval are inserted in the respective FIFOs 66, 68 (or in separate internal FIFO registers of a single FIFO in an alternative embodiment). In the following clock period, the functions of time slots A and B are swapped. In this way, no pulses generated by the Mini-SiPM 40a, 40b, 40c, 40d are lost during data transfer and zeroing.

How does the FIFO 66, 68 work? described in the previous paragraph with reference to the mode? inactive. In this mode, the FIFOs ensure that detection signals, including photon counts, and time stamp information are retained for several clock periods after they are measured. Does the FIFO function change when the chip enters the mode? event monitoring in response to the combined detection signal from all pixels exceeding the threshold (s) and the fact that an event? therefore detected by the discriminator 22.

When an event is detected, the discriminator 22 generates an activation and goes into the mode? of integration. The activation signal, which can? be referred to as a blocking signal,? distributed to the processing logic of each of the pixels 20, and in response each pixel enters the mode? event monitoring, which means that each pixel begins to accumulate counts, while the FIFOs of the TDC 66 will be blocked. The FIFOs 66, 68 must keep the values stored in the FIFOs 66, 68 in the time interval in which? the event was detected, considering the latency to generate and propagate the activation. The FIFO compensate cos? the latency of passage through the adder shaft and the discriminator. FIFOs ensure that all data relating to a particular event are preserved (in this case both the number of photons and time stamp data), and the integration begins with the interval values that caused the activation of discriminator 22, ensuring that all photon counts associated with a particular event are used, even if they appear at the beginning of the event.

When ? controlled by the activation signal, distributed in synchronism with the clock, the pixel processing logic, in the form of the DAMAC 42, accumulates the pulses of the SPADs counted in each clock cycle during the integration period, using the accumulator (which in this case it is part of FIFO 68, although it can be prepared in the form of a separate component). How will it be? described, each block 52, 54 of Mini-SiPM Interface and Counter ("Mini-SiPM Interface and Counter" - MIC) 52, 54 contains an OR shaft to combine the outputs of the single SPADs in a single sequence of pulses, with monostable devices after the first OR stage which perform a function of shortening the pulses to prevent overlapping pulses from obscuring each other. The counter 54, also in the MIC block, adds up the pulses of the SPADs generated during each clock cycle, made using the pair of 'ripple' counters operating in one mode. alternating action, so that one is integrating while the other is being read and reset. The pixel-level DA-MAC 42 then adds the counter outputs of the four Mini-SiPMs, and the accumulator does the integration.

Finally, when the end of the programmable integration time (IntTime) is reached, the accumulator is also blocked and the integrated counts for each pixel and the TDC values of both the first and the second integrated interval are read by the discriminator 22.

Alternatively, the internal discriminator 22 can? be bypassed in favor of an external discriminator or other processing resource, such as a central processor, since? both integration and reading signals can be supplied externally.

When read, the accumulator values of the individual pixels provide an 'image' of intensity. showing the spatial distribution of scintillation light across the sensor, allowing the scintillation point to be determined by off-chip processing. Similarly, the values can be added to calculate the overall energy of the incident gamma photon. As indicated above, the TDCs in the pixels are arranged in an alternating action pair, allowing a single time stamp to be generated for each clock cycle without an intermediate dead time. These CCT values are placed in their FIFO 66 which? locked at the start of integration, retaining the time stamp of the crucial first photons in the scintillation pulse, ready to be read along with the spatial data as already? mentioned above.

In addition to obtaining the intensity image, the discriminator 22, or an external discriminator off the chip or other processing resource, can be used. also continue, during the mode? event monitoring, to receive the combined detection signal which represents the sum of detection signals (in this case photon numbers) which? was used to initially determine the occurrence of the event. So, a temporal histogram of the counts can? be monitored externally in real time, and an analysis of these data can? allow the filtering of false events and the detection of multiple simultaneous gamma events or other events. This analysis can? it can also be used to determine a type of occurrence, or other property, of an event, for example by interpolating the data on the basis of an appropriate model.

A time chart, and a histogram of total photon counts obtained before and during a detection event, are provided in Figure 6 for illustrative purposes. Can you? see that the determination of the occurrence of a detection event, shown in the diagram as a validation decision, occurs some time after the start of the pulse. However, the use of FIFOs, and the continuous output of the combined detection signal, ensures that all relevant pixel count data can be used by embedding it in the event histogram. An additional detailed timeline? provided in Figure 7 for reference.

In the embodiment illustrated in FIGS. 2 to 5, the processing of detection signals and time stamp data associated with a detection event can? be performed off-chip, such as in a central processor or other resource, with relevant data provided by the chip over a communication bus. In alternative embodiments, this processing or any desired part of this processing can be done. be performed by a processor disposed on the chip, and the results of such processing can be output from the chip as desired. Examples of such processing include, but are not limited to, averaging or other combined processing of time stamp data for all pixels to determine the time at which it is? verified the event, the analysis of pixel count profiles as a function of time, the processing of the total count data for each pixel to obtain image data and / or to determine the position in which it? event occurred.

A characteristic of the embodiment illustrated in Figures 2 to 5 consists in the fact that the components of the processing logic of each pixel are positioned in a logic area of the pixel arranged in a central position, such as? schematically illustrated in Figure 8b. In alternate embodiments, the individual SPADs may be surrounded by SPADs as illustrated in Figure 8a, but the arrangement illustrated in Figure 8b may? allow a better fill factor. The fill factor can? also be enhanced by using a cathode well as illustrated in Figure 9, which shows (a) a plan view, and (b) a cross-sectional view representing individual SPADs surrounded by a logic in a 130 nm CMOS image generation technology, and in Figure 10, showing (a) the arrangement of a shared SiPM with logic channel, and (b) a cross-sectional view showing the structure of the device. The arrangement illustrated in Figures 8b to 10 allows to realize 720 (4? 180) SPADs in a compact pixel area of 0.57? 0.61 mm2 together with the associated electronics. However, care must be taken, in low voltage CMOS technologies, to ensure that the high voltage applied to the NWELL cathode does not cause parasitic junctions to fail, which? resolved herein with the use of an undoped "EPI" protection ring to prevent lateral discharge as illustrated in FIG. 11a. Other approaches that can be used in alternative embodiments are illustrated in Figures 11b and 11c.

The sensor of the embodiment illustrated in Figures 2 to 5? been verified with a Teflon-coated LYSO scintillator crystal of 3? 3? 5 mm irradiated from a 370 kBq, 511 keV Na22 gamma source. The system provided an energy resolution of 13.1% FWHM ("Full Width at Half Maximum"? Width at half? Height) at 20? C, how? shown in Figure 12, and an improved FWHM of 11.3% with cooling to -20 ° C and using a higher polarization of the SPADs of 14.5 V. A 70 ps pulsed laser was used to measure the temporal resolution of the sensor, operated at a low emission power with a diffuser to reduce saturation effects ("pile-up" in the Anglo-Saxon technical literature).

Figure 13 shows a plot of TDC codes generated from a single pixel, showing an FWHM of 263 ps, compared to a reference system. When averaging all valid TDC codes in a frame, the resulting FWHM? improved to 239 ps, demonstrating that the per-pixel TDC architecture provides better time performance than per-column or per-sensor time stamp. The device can? include, in some embodiments, a processing resource for processing obtained timestamp data for all pixels in order to obtain a higher determination. details of the instant in which an event occurs compared to that which could be obtained from time stamp data for a single pixel. Is it possible to perform any appropriate combined processing, and such processing is not? limited to averaging.

The embodiment illustrated in Figures 2 to 5 can allow the integration of the sum of photons incident on the entire surface of the sensor in short time intervals (defined by a clock waveform), and the availability? continuous of this sum continuously at the sensor output. The embodiment described can? providing time stamps for each pixel, which are placed in FIFO buffers for each pixel allowing access to the time history of successive photon arrivals.

The embodiment illustrated in Figures 2 to 5 can allow the determination of the time, energy and spatial position of events through the sensor architecture consisting of a pixel matrix and an acquisition with a high degree of parallelism of the number of photons detected and the arrival time of these photons ( with multiple CCT). Such a matrix can? operate as a single large area detector where the total number of photons detected? available in real time at the top level through the network of distributed adders.

The embodiment illustrated in Figures 2 to 5 can also allow to distinguish photonic events (such as gamma events) from noise events. The SPADs operate continuously and asynchronously (the SPADs are self-reloading) and the sensor? able to count continuously the photons in time intervals which are used for the validation of photonic events. The architecture of the sensor? able to provide, for example, a global number of photons detected for each interval. What? the validation of a possible event as a genuine event of a selected type (for example a gamma event) can? be performed using complex algorithms based on, or taking into account, the temporal evolution of the counts.

Bench? the sensor device according to the embodiment illustrated in Figures 2 to 5 has been described with reference to measurements for PET, the sensor device according to alternative embodiments is not? limited to such measures. For example, alternative embodiments can be set up for measurements for MRI or for any asynchronous detection of photonic events, such as in particle physics, X-ray analysis or spectrometry, where time, energy and / or the spatial position of events.

The sensor device does not? limited to the embodiment described with reference to Figures 2 to 5, and it will be understood? for example that various aspects of the functionality? of the embodiment illustrated in Figures 2 to 5 can be implemented without other aspects of functionality? in alternative embodiments. For example, in some alternative embodiments,? Is it possible to foresee the temporary storage of pixel detection signals in times before the identification of the occurrence of an event, for example by using one or more? FIFO of pixels, but? It is possible not to provide a hierarchical tree or other structure for the combination of signals, and vice versa.

In alternative embodiments,? It is possible to provide any other suitable data storage means for storing pixel detection signals, in addition to, or in place of, pixel FIFOs. Similarly, in alternative embodiments,? It is possible to provide any other appropriate arrangement for combining different pixel detection signals, in addition to, or in place of, a hierarchical structure of H-tree adders. For example, in some alternative embodiments,? It is possible to use any suitable tree or other structure which combines signals corresponding to the same time interval and / or which provides substantially equal routing delays between nodes. Alternatively, in some embodiments, the combination of pixel detection signals in the same time interval can? be performed by a processing resource based on received time interval data associated with each pixel detection signal, although? there? may be less efficient than using a distributed adder tree.

In alternative embodiments, the functionality? of the event monitoring device to identify the occurrence of an event can? be supplied, for example, by a single component or by a plurality of of distributed components, and such components can be arranged locally, for example on the same chip as the pixels, or remotely, for example as part of a central processing resource.

Will you understand? that the present invention? previously described purely by way of example, and that various modifications of detail can be made within the scope of the invention.

Each feature depicted in the description and (where appropriate) in the claims and drawings can? be provided independently or in any appropriate combination.

Claims (10)

CLAIMS 1. Sensor device for photon detection, comprising: a plurality? of pixels, with each pixel comprising a plurality of pixels of photonic detectors and a processing logic adapted to output a respective detection signal for each pixel for each interval of a series of time intervals; a combiner adapted, for each time interval, to combine the detection signal of each pixel to produce a combined detection signal for this time interval, an event monitoring device adapted to identify the occurrence of a detection event during a particular time interval as a function of a value of the combined detection signal for at least this time interval. 2. Device according to claim 1, wherein the sensor device operates in accordance with a first mode? before identifying a detection event, and? able to operate in accordance with a second modality? different in response to the event monitoring device having identified the occurrence of a detection event. A device according to claim 2, further comprising means for obtaining an integrated detection signal for each pixel representative of accumulated detection signals during an integration period in the second mode. Device according to any one of the preceding claims, wherein, for each pixel, the processing logic comprises at least one time stamp device which? adapted to provide a time stamp to record the instant of detection of one or more? photons by the detectors of this pixel, in which the time stamp optionally records the instant of detection with a resolution pi? end of the length of the time interval; and wherein the device further comprises means for determining a time of the event from time stamp data obtained from each pixel of the plurality. of pixels, for example by calculating the average of time stamp data obtained from the plurality? of pixels. 5. Device according to any one of the preceding claims, further comprising, for each pixel, a data storage medium which? adapted to store at least one detection signal and / or at least one time stamp for at least one previous time interval for this pixel; And in which, for each pixel, the data storage medium? capable of storing, upon identification of the occurrence of a detection event, at least one detection signal and / or at least one time stamp for at least a time interval preceding, during, or following the identification of the occurrence of the event detection. Device according to claim 5, wherein, for each pixel, the data storage medium comprises at least one FIFO for storing the at least one detection signal and / or at least one FIFO for storing the at least one time stamp. Device according to any one of claims 2 to 6, wherein the event monitoring device? able to send one or more? control signals to the processing logic of each pixel in response to the identification of a detection event, to block the operation of the at least one FIFO of this pixel while in the second mode. and / or to control the integrator for this pixel to accumulate the integrated detection signal for the integration period. 8. Device according to any one of the preceding claims, wherein the combiner comprises a tree structure comprising a plurality of elements. of nodes in a plurality? of levels, in which the pixel processing logic provides pixel detection signals to the root nodes of the tree and the tree structure? configured so that, in each tiered node above the root node, signals from at least two nodes at a lower level of the tree are combined. Device according to any one of the preceding claims, wherein the combiner comprises a hierarchical tree structure of adders. 10. Device according to any one of the preceding claims, further comprising means for obtaining, for each detection event, one or more, optionally each, of: a profile of the combined detection signal for all pixels as a function of time during the integration period; an integrated detection signal for each pixel, representative of the detection signals accumulated by this pixel during the integration period; time stamp data for each pixel.