WO2023082307A1

WO2023082307A1 - Method and apparatus for calculating depth of interaction, and computer-readable storage medium

Info

Publication number: WO2023082307A1
Application number: PCT/CN2021/131636
Authority: WO
Inventors: 程冉; 肖鹏; 汪飞
Original assignee: 苏州瑞派宁科技有限公司
Priority date: 2021-11-12
Filing date: 2021-11-19
Publication date: 2023-05-19
Also published as: CN114186166A

Abstract

Disclosed in the present application are a method and apparatus for calculating a depth of interaction, and a computer-readable storage medium. The method comprises: acquiring, from a photoelectric sensor/a photoelectric sensor array, feature data related to photons; and inputting the feature data into a preset ensemble learning model, so as to calculate the depth of interaction of the photons. In some embodiments of the present application, the solution is simple and has a high accuracy, and can also be implemented more conveniently by using an FPGA and be popularized and used more easily.

Description

Method, device and computer-readable storage medium for calculating reaction depth

technical field

The present application relates to the field of data processing, in particular, to a method, device and computer-readable storage medium for calculating reaction depth.

Background technique

The response line in the PET (Position Emission Tomography, referred to as PET) system is a pair of detectors that receive gamma photons at the same time. Since the crystal has a certain length, for example, the small animal PET is 13mm, and the clinical PET is 20mm, so that the gamma photon has a certain reaction depth, that is, DOI (Depth of Interaction, referred to as DOI) information.

When there is no DOI information, the connection line of the response line can only be determined as the connection line of the center of the detector surface, and this will cause the positioning error of the response line, cause parallax effect, and affect the spatial resolution of the system and the shape of the image. As shown in Figures 1a-1c, due to the parallax effect caused by the uncertain depth of gamma photons deposited in the crystal, the resolution of the PET image is low and the resolution is not uniform. Especially in four-plate PET, the parallax effect between two adjacent plates is more obvious. A large number of research results show that the radial resolution will gradually deteriorate from the center of FOV (Field of View, FOV for short) to the edge of FOV, seriously affecting the image resolution, so the measurement of DOI is very important to improve the resolution of PET images.

Currently, for pixelated single crystal strips, that is, array crystals, the DOI is mainly calculated by double-end readout or single-end readout.

The original double-ended readout method is to use different optical sensors at both ends of the array crystal, such as one end coupled with a PMT, and one end coupled with a photodiode, and measure the gamma photon by comparing the energy of the signal detected by the optical sensors at both ends. Depth of reaction in the crystal. Now, it is often used to couple SiPM (Silicon Photo-Multiplier, SiPM) at both ends of the array crystal, and process the four sides of the crystal, for example, rough treatment, adding a reflective layer, etc., and finally according to the energy of one end and the two ends The ratio of the energy sum to determine the DOI position.

For the single-ended readout method, the windowed light sharing method is currently used more often. This method mainly uses the light distribution of the crystal on the SiPM to establish the relationship with the DOI. As shown in Figure 2, crystal No. 3 and crystal No. 4 have a light-sharing window at the white position in the figure. Gamma photons enter crystal 3. If they are deposited at position C, a large amount of visible light will propagate into crystal 4 and be detected by SiPM2 However, SiPM1 detects very little; if it is deposited at A position, a large amount of visible light will only propagate into crystal 3 and be detected by SiPM1, while SiPM2 is hardly detected. At this point, the light distribution on the SiPM array is no longer a point, but elongated, as shown in Figure 3a and Figure 3b. DOI can be obtained by establishing the transfer function of light distribution and DOI.

Light distribution refers to photon counts on individual SiPMs in the SiPM array. As shown in Figure 4a and Figure 4b, a gamma photon into the crystal array will have a light distribution on the coupled SiPM array after deposition at a certain position. The shallower the deposition position (that is, the reaction depth), the light distribution The wider the range; the deeper the deposition position, the more concentrated the light distribution.

There are some disadvantages in the prior art method of calculating DOI, for example, the double-ended readout method will double the number of SiPMs and corresponding circuits, and the SiPMs and circuits close to one end of the FOV will affect the detection of gamma photons. However, the windowed light sharing method will cause a small number of crystals to have no DOI capability due to the difference in windowing.

The prior art method for calculating the DOI increases the complexity of the system design to a certain extent, requires additional space, and consumes a lot of power. For highly integrated small animal PET or PET/MR, increased power consumption and extra space are not conducive to cost savings and optimized system design.

Contents of the invention

The present application provides a method, device and computer-readable storage medium for calculating reaction depth, which solve at least one of the above-mentioned problems.

According to one aspect of the present application, a method for calculating the depth of reaction is proposed, the method comprising: obtaining characteristic data related to photons from a photoelectric sensor/photoelectric sensor array; inputting the characteristic data into a preset integrated learning model , to calculate the photon response depth.

According to some embodiments of the present application, the method further includes: using sample data to train the integrated learning model.

According to some embodiments of the present application, the sample data includes feature data obtained from one or both ends of the photosensor/photosensor array.

According to some embodiments of the present application, the sample data includes characteristic data of the reaction depth and the reaction depth.

According to some embodiments of the present application, the method further includes: combining multiple ensemble learning sub-models to form the ensemble learning model.

According to some embodiments of the present application, the inputting the feature data into a preset integrated learning model to calculate the reaction depth of photons includes: using the feature data in each of the integrated learning sub-models Performing calculations to generate photon reaction depths respectively; according to the calculation results of the plurality of integrated learning sub-models, calculating the photon reaction depths by voting or calculating an average value.

According to some embodiments of the present application, the integrated learning model is calculated using one or more of a random forest algorithm, a boosted tree algorithm, or a gradient boosted tree algorithm.

According to some embodiments of the present application, the feature data includes energy information corresponding to photons, pulse sampling point information, pulse rising edge time, pulse energy, pulse decay time and/or light distribution on the photosensor/photosensor array.

According to some embodiments of the present application, the acquiring photon-related characteristic data from the photosensor/photosensor array includes: obtaining the photosensor/photosensor coupled to one or both ends of the scintillation crystal/scintillation crystal array An array acquires the feature data.

According to some embodiments of the present application, the photoelectric sensor adopts PMT and SiPM; when the two ends of the scintillation crystal/scintillation crystal array are coupled to the photoelectric sensor/photoelectric sensor array, PMT or SiPM is set at both ends, or a PMT is set at one end , set SiPM at one end.

According to some embodiments of the present application, the photons include high-energy photons among X-rays, γ-rays, α-rays, β-rays, and neutron rays.

According to one aspect of the present application, a device for calculating the depth of reaction is proposed, the device includes: a data acquisition unit for obtaining characteristic data related to photons from a photoelectric sensor/photoelectric sensor array; a reaction depth calculation unit for The feature data is input into a preset integrated learning model to calculate the response depth of the photon.

According to one aspect of the present application, a device for calculating the reaction depth of gamma photons is proposed, including one or more processors; a storage device for storing computer programs; when the computer program is processed by the one or more When executed by a processor, the one or more processors implement the method as described in any one of the preceding ones.

According to one aspect of the present application, a computer-readable storage medium is provided, on which program instructions are stored, and when the program instructions are executed, the method described in any one of the preceding items is implemented.

According to some exemplary embodiments of the present application, using feature data related to photons to estimate the reaction depth of photon deposition positions in the crystal through an integrated learning model is simple and easy to implement, and the accuracy rate is also higher. And because the integrated learning model is simple to implement, it is easier to use FPGA to implement, and it is easier to promote and use.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings that need to be used in the description of the embodiments.

Figure 1a shows a schematic diagram of response line relocation based on photon deposition coordinates obtained from simulation.

Figure 1b shows the reconstructed pattern when the reaction depth is 2mm.

Figure 1c shows the reconstructed image without reflecting depth information.

Fig. 2 shows a schematic cross-sectional view of a detector using a single-ended readout method.

Figure 3a shows the light distribution diagram of the SiPM array.

Figure 3b shows a schematic diagram of photon counting comparison on a SiPM array.

Figure 4a shows a schematic diagram of the light distribution after gamma photons are injected into the continuous crystal deposition.

Fig. 4b shows a schematic plan view of the light distribution on the coupled SiPM array after the gamma photons are injected into the continuous crystal.

Fig. 5 shows a flowchart of a method for calculating reaction depth according to an exemplary embodiment of the present application.

Figure 6 shows a decision tree structure.

Fig. 7a shows a random forest model according to an exemplary embodiment of the present application.

Fig. 7b shows a schematic diagram of estimating reaction depth by using a random forest model according to an exemplary embodiment of the present application.

Fig. 8 shows a flowchart of a method for training an ensemble learning model according to an exemplary embodiment of the present application.

Fig. 9 shows a screenshot of a method for acquiring sample data according to an exemplary embodiment of the present application.

Fig. 10 shows a block diagram of an apparatus for calculating reaction depth according to an exemplary embodiment of the present application.

Fig. 11 shows a block diagram of another device for calculating reaction depth according to an embodiment of the present application.

Detailed ways

Example embodiments will now be described more fully with reference to the accompanying drawings. Example embodiments may, however, be embodied in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the concept of example embodiments to those skilled in the art. The same reference numerals denote the same or similar parts in the drawings, and thus their repeated descriptions will be omitted.

The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided in order to give a thorough understanding of embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced without one or more of these specific details, or other methods, components, materials, devices or operations may be used. In these instances, well-known structures, methods, devices, implementations, materials, or operations are not shown or described in detail.

The flow charts shown in the drawings are only exemplary illustrations, and do not necessarily include all contents and operations/steps, nor must they be performed in the order described. For example, some operations/steps can be decomposed, and some operations/steps can be combined or partly combined, so the actual order of execution may be changed according to the actual situation.

The terms "first", "second" and the like in the description and claims of the present application and the above drawings are used to distinguish different objects, rather than to describe a specific order. Furthermore, the terms "comprising" and "having", as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes unlisted steps or units, or optionally further includes For other steps or units inherent in these processes, methods, products or devices.

Specific embodiments according to the present application will be described in detail below with reference to the accompanying drawings.

Fig. 5 shows a flowchart of a method for calculating reaction depth according to an exemplary embodiment of the present application. A method for calculating the reaction depth according to an exemplary embodiment of the present application will be described in detail below with reference to FIG. 5 .

In step S501, feature data related to photons is acquired from the photoelectric converter/photosensor array.

According to some embodiments of the present application, the characteristic data related to photons are obtained through the photoelectric converter/photosensor array in the detector, and the detector includes a scintillation crystal/scintillation crystal array and a photoelectric sensor coupled to at least one end of the scintillation crystal/scintillation crystal array. A converter/photoelectric converter array, wherein the scintillation crystal converts photons into visible light, the photoelectric converter converts visible light into electrical signals, and characteristic data related to photons is obtained from the electrical signals.

According to some embodiments of the present application, the photoelectric converter array includes PMT, PMT array, SiPM, SiPM array, and a single element in the photoelectric converter array can communicate with crystal elements in the scintillation crystal array in a one-to-one or one-to-many manner. coupling. When photoelectric converters are installed at both ends of the scintillation crystal, the types of photoelectric converters at both ends can be the same or different, for example, both ends use PMT/SiPM, or one end uses PMT and the other end uses SiPM. Although the performance of PMT is inferior to that of SiPM, in some specific applications, such as applications with low imaging quality requirements, using PMT is more cost-effective than using SiPM.

According to some example embodiments of the present application, the photons may be any high-energy photons capable of photoelectric conversion in the scintillation crystal, such as high-energy photons in X-rays, γ-rays, α-rays, β-rays, neutron rays, and the like.

According to some exemplary embodiments of the present application, the acquired photon-related feature data includes gamma photon energy information, pulse sampling point information, pulse rising edge time, pulse energy, pulse decay time, and/or light on the photosensor array. Distribution, for example, for a gamma photon, its characteristic energy is usually 511keV, and the converted electrical pulse signal usually includes a relatively fast rising edge and a relatively slow falling edge.

According to some embodiments of the present application, characteristic data related to photons is obtained from electrical signals output by photosensor arrays at one or both ends of the scintillation crystal array.

In step S503, input the photon-related feature data obtained in step S501 into a preset integrated learning model to calculate the response depth of gamma photons.

According to some embodiments of the present application, the integrated learning model is calculated by using one or more of random forest algorithm, boosting tree algorithm or gradient boosting tree algorithm.

According to some embodiments of the present application, the integrated learning model in step S503 includes a plurality of integrated learning sub-models, and is obtained by training with sample data. Wherein, the sample data includes the characteristic data of the reaction depth and the reaction depth. When step S503 is executed, according to the calculation results of multiple integrated learning sub-models, the reaction depth of the photon is calculated by voting or calculating an average value.

Taking the random forest model in the random forest algorithm as an example, since the random forest uses multiple decision trees to achieve the goal, each decision tree will give a value. When step S503 is executed, the photon-related feature data is input into the random forest model, and each branch of the random forest model will obtain an estimated value of the response depth. Therefore, the final result is finally obtained by voting or averaging .

According to some embodiments, the integrated learning model in step S503 can also select various types of integrated learning models, such as boosted tree model and gradient boosted tree model, and finally determine the final response depth value by voting or averaging.

According to some embodiments, if the feature data in step S501 is obtained through a single-end photoelectric converter array, then the integrated learning model in step S503 is also trained through single-end sample feature data.

According to some embodiments, if the feature data in step S501 is obtained through two-terminal photoelectric converter arrays, then the integrated learning model in step S503 is also trained through two-terminal sample feature data.

Taking the random forest in the ensemble learning model as an example, how to calculate the response depth of photons according to the embodiment shown in FIG. 5 will be described in detail below.

Random forest is one of the integrated learning algorithms. Random forest uses multiple decision trees to achieve the goal. Each decision tree will give a value, and finally the final result is obtained by voting or averaging. Fig. 6 shows a decision tree structure, where each internal node represents a test on an attribute, each branch represents a test output, and each leaf node represents a category or predicted value.

The energy data in the gamma photon-related feature data and the decision tree shown in FIG. 7a and FIG. 7b are used below to illustrate how to estimate the reaction depth of photons according to the embodiment shown in FIG. 5 .

As shown in Figure 7a and Figure 7b, each node of the decision tree is a judgment condition. Assuming that both ends of the scintillation crystal strip/array are coupled to the photoelectric converter/array, the input energy at both ends of a single crystal strip is 4476.8 for CH1 Energy and 6825.6 for CH2 Energy. According to the decision tree described in Figure 7a, the first layer When judging, because CH1 Energy is 4476.8 less than 5594.0, it meets the judgment condition "CH1 Energy ≤ 5594.0" of the first layer node, so it enters the branch node on the left side of the second layer. In the same way, judge the branch direction of the second layer. Since CH2 Energy is 6825.6, which does not meet the judgment condition of this node "CH2 Energy ≤ 6525.2", it enters the branch node on the right side of the next level, and finally obtains the black curve in Figure 7b According to the judgment route, the value of the reaction depth is 5.105mm.

The embodiment shown in Fig. 7a and Fig. 7b is based on the energy data in the characteristic data related to the gamma photon. Similarly, the characteristic data such as sampling point and pulse time can also be used to estimate the reaction depth at the same time, and according to each characteristic data, the The final reaction depth is calculated by voting or averaging. The more feature data used, the more accurate the depth of response obtained.

In the embodiment shown in Figure 7a and Figure 7b, the judgment conditions of each node of the decision tree and the value of the reaction depth at the leaf node are trained by sample data, random input data randomly selects features, and the input samples of each tree are not necessarily all samples, so it is not prone to overfitting. Since it is composed of different trees, it can be used to process nonlinear data and fit nonlinear models, which can make the judgment conditions and leaf node values obtained closer to the real values.

According to the above example embodiments, by using the photon-related feature data and estimating the deposition position of the photon in the crystal through the integrated learning model, the DOI information can be obtained using only single-ended data, avoiding the back-end caused by the double-ended readout detector. The change of the circuit to the existing system structure effectively solves the problem that the photon detection is affected by the front-end photoelectric conversion device and circuit in the double-end readout mode. In addition, at the system level, cost savings will not cause additional load caused by increasing the number of channels, and avoid problems such as gaps caused by system structure changes.

At present, the average absolute error of the DOI estimation of the 3mm×3mm×20mm scintillation crystal array calculated by the above-mentioned embodiment is about 1mm, which is higher in accuracy than the double-ended readout method in the prior art.

The above integrated learning model is simple to implement, easier to implement with FPGA, and easier to promote and use in real systems.

Fig. 8 shows a flowchart of a method for training an ensemble learning model according to an example embodiment of the present application. A method for training an ensemble learning model according to an exemplary embodiment of the present application will be described in detail below with reference to FIG. 8 .

As shown in FIG. 8 , in step S801 , an ensemble learning model type is selected.

According to some embodiments of the present application, the integrated learning model is calculated by using a random forest algorithm, a boosted tree algorithm or a gradient boosted tree algorithm.

In step S803, sample data is acquired.

Fig. 9 shows a method of acquiring sample data. As shown in Figure 9, a source Source is placed on the side of a crystal with a height of 20mm. The upper and lower ends of the crystal are coupled with SiPM arrays. From a specific position on the side of the crystal, such as 30cm, gamma photons are input and pass through the other side. The collimating crystal monitors whether the incident gamma photons are collimated, and then moves the incident position in turn to obtain the pulse signal at both ends at a known DOI position, that is, at a specific depth, and then obtain the characteristic data of the pulse signal, such as energy signal etc. It should be noted by those skilled in the art that the above placement locations and distances are merely examples and not limiting, and those skilled in the art can select appropriate locations and distances for data acquisition as required.

In step S805, the integrated learning model is trained.

According to an exemplary embodiment of the present application, step S803 is used to obtain feature data at both ends and corresponding DOI position data to train an integrated learning model.

According to some embodiments, if an integrated learning model at one end is required, when collecting feature data, only the pulse signal feature data at one end needs to be collected, and the integrated learning model is trained using the pulse signal feature data at one end and the corresponding DOI position data.

Fig. 10 shows a block diagram of an apparatus for calculating reaction depth according to an exemplary embodiment of the present application. As shown in FIG. 10 , a device for calculating the reaction depth includes a data acquisition unit 1001 and a reaction depth calculation unit 1003 . Wherein, the data acquisition unit 1001 is used to acquire photon-related feature data from the photoelectric sensor/photosensor array, and the response depth calculation unit 1003 is used to input the feature data into a preset integrated learning model to calculate the photon response depth.

Fig. 11 shows a block diagram of another device for calculating reaction depth according to an embodiment of the present application. The device for calculating the reaction depth shown in FIG. 11 is only an example, and should not limit the function and application scope of the embodiment of the present application.

As shown in Fig. 11, the means for calculating the depth of reaction is represented in the form of a general-purpose computing device. The components of the apparatus for calculating reaction depth may include but not limited to: at least one processor 210, at least one memory 220, a bus 230 connecting different system components (including memory 220 and processor 210), a display unit 240, and the like. Wherein, the memory 220 stores program codes, and the program codes can be executed by the processor 210, so that the processor 210 executes the methods described in this specification according to various exemplary embodiments of the present application. For example, the processor 210 may execute the method as shown in FIG. 5 .

The memory 220 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 2201 and/or a cache storage unit 2202 , and may further include a read-only storage unit (ROM) 2203 .

Memory 220 may also include programs/utilities 2204 having a set (at least one) of program modules 2205 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, which Each or some combination of the examples may include the implementation of a network environment.

Bus 230 may represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local area using any of a variety of bus structures. bus.

The means for calculating the depth of reaction may also communicate with one or more external devices 300 (such as keyboards, pointing devices, bluetooth devices, etc.), and may also communicate with one or more devices enabling the user to interact with the means for calculating the depth of reaction, And/or communicate with any device (eg, router, modem, etc.) that enables the means for calculating the photon's depth of response to communicate with one or more other computing devices. Such communication may occur through input/output (I/O) interface 250 . Furthermore, the means for calculating the depth of response can also communicate with one or more networks (eg, local area network (LAN), wide area network (WAN) and/or a public network, such as the Internet) via network adapter 260 . Network adapter 260 may communicate with other modules of the apparatus for calculating depth of response via bus 230 . It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the means for calculating response depth, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, Tape drives and data backup storage systems, etc.

Through the description of the above implementations, those skilled in the art can easily understand that the example implementations described here can be implemented by software, or by combining software with necessary hardware. The technical solutions according to the embodiments of the present application can be embodied in the form of software products, which can be stored in a computer-readable storage medium (which can be CD-ROM, U disk, mobile hard disk, etc.) or on the network, including several The computer program instructions enable a computing device (which may be a personal computer, a server, or a network device, etc.) to execute the above method according to the embodiments of the present application.

A software product may utilize any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: electrical connection with one or more conductors, portable disk, hard disk, random access memory (RAM), read only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

A computer readable storage medium may include a data signal carrying readable program code in baseband or as part of a carrier wave traveling as part of a data signal. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A readable storage medium may also be any readable medium other than a readable storage medium that can send, propagate or transport a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the readable storage medium may be transmitted by any suitable medium, including but not limited to wireless, cable, optical cable, RF, etc., or any suitable combination of the above.

Program codes for performing the operations of the present application can be written in any combination of one or more programming languages, including object-oriented programming languages—such as Java, C++, etc., as well as conventional procedural programming Language—such as C or a similar programming language. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server to execute. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (for example, using an Internet service provider). business to connect via the Internet).

The above-mentioned computer-readable medium carries one or more program instructions, and when the above-mentioned one or more program instructions are executed by one device, the computer-readable medium can realize the above-mentioned functions.

Those skilled in the art can understand that the above modules can be distributed in the device according to the description of the embodiment, and can also be changed correspondingly to one or more devices that are only different from the embodiment. Multiple modules in the above embodiments may be combined into one module, or one module may be further split into multiple sub-modules.

Those skilled in the art can understand that the above-mentioned modules can be distributed in the device according to the description of the embodiment, and corresponding changes can also be made in one or more devices that are only different from the embodiment. Multiple modules in the above embodiments may be combined into one module, or one module may be further split into multiple sub-modules.

According to some example embodiments of the present application, using the photon-related feature data, the reaction depth is estimated for the photon deposition position in the crystal through an integrated learning model, and it is realized that only using The ability to obtain DOI from single-ended data avoids the change of double-ended readout detectors and circuits to the existing system structure, as well as the impact on photon detection, and the accuracy is also higher. And because the integrated learning model is simple to implement, it is easier to use FPGA to implement, and it is easier to promote and use.

Although the present application provides the operation steps of the method described in the above embodiments or flowcharts, more or fewer operation steps may be included in the method based on routine or no creative effort. In the steps where logically there is no necessary causal relationship, the execution order of these steps is not limited to the execution order provided in the embodiment of the present application.

Each embodiment in this specification is described in a progressive manner, the same and similar parts of each embodiment can be referred to each other, and each embodiment focuses on the differences from other embodiments.

The above is a detailed introduction to the embodiments of the present application. In this paper, specific examples are used to illustrate the principles and implementation methods of the present application. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present application. At the same time, changes or deformations made by those skilled in the art based on the ideas of the application, specific implementation methods and application scopes of the application all belong to the scope of protection of the application. To sum up, the contents of this specification should not be understood as limiting the application.

Claims

A method for calculating the depth of reaction, characterized in that the method comprises:

Obtain photon-related characteristic data from photosensors/photosensor arrays;

The feature data is input into a preset integrated learning model to calculate the response depth of the photon.
The method according to claim 1, further comprising: using sample data to train the integrated learning model.
The method according to claim 2, wherein the sample data includes characteristic data obtained from one or both ends of the photosensor/photosensor array.
The method according to claim 2, wherein the sample data includes characteristic data and reaction depth of the reaction depth.
The method according to claim 1, further comprising: combining a plurality of ensemble learning sub-models to form the ensemble learning model.
The method according to claim 5, wherein said inputting said feature data into a preset integrated learning model to calculate the reaction depth of photons comprises:

In each of the integrated learning sub-models, the characteristic data are used to perform calculations to generate the response depths of photons respectively;

According to the calculation results of multiple integrated learning sub-models, the reaction depth of the photon is calculated by voting or calculating the average value.
The method according to claim 1, wherein the integrated learning model is calculated using one or more of a random forest algorithm, a boosted tree algorithm, or a gradient boosted tree algorithm.
The method according to claim 1, wherein the feature data includes photon-corresponding energy information, pulse sampling point information, pulse rising edge time, pulse energy, pulse decay time and/or photoelectric sensor/photoelectric sensor array light distribution.
The method according to claim 1, wherein said acquiring photon-related characteristic data from the photoelectric sensor/photoelectric sensor array comprises: The photosensor/photosensor array acquires the characteristic data.
The method according to claim 9, wherein the photoelectric sensor adopts PMT or SiPM; when the two ends of the scintillation crystal/scintillation crystal array are coupled to the photoelectric sensor/photoelectric sensor array, PMT or SiPM are set at both ends , or set a PMT at one end and a SiPM at one end.
The method according to claim 1, wherein the photons include high-energy photons among X-rays, γ-rays, α-rays, β-rays, and neutron rays.
A device for calculating the depth of reaction, characterized in that the device comprises:

The data acquisition unit is used to obtain characteristic data related to photons from the photoelectric sensor/photoelectric sensor array;

The reaction depth calculation unit is used to input the characteristic data into the preset integrated learning model to calculate the reaction depth of the photon.
A device for calculating the depth of reaction, characterized in that it comprises:

one or more processors;

storage means for storing computer programs;

When the computer program is executed by the one or more processors, the one or more processors are made to implement the method according to any one of claims 1-11.
A computer-readable storage medium, characterized in that program instructions are stored thereon, and the method according to any one of claims 1-11 is implemented when the program instructions are executed.