Intelligent positioning method of incident rays based on continuous crystal gamma imaging detector

Abstract

The present invention discloses an incident ray intelligent positioning method based on a continuous crystal gamma imaging detector, wherein the continuous crystal gamma imaging detector includes a detector and an ADC, wherein the gamma ray beam acts on the incident surface of the detector and is converted into a pulse signal array, and then converted into a response matrix by the ADC. The present invention first establishes a coordinate system on the incident surface of the detector according to the size of the incident surface of the detector, determines the scanning point, the two-dimensional coordinates of the scanning point, and the scanning path; then scans point by point, obtains the response matrix corresponding to the full energy peak count under each scanning point, processes it into data samples to form a data set, and then uses the KNN algorithm to locate the rays emitted by the radiation source at an unknown position. The present invention belongs to the field of artificial intelligence and gamma ray imaging, and can improve the accuracy of the detector in locating the position of the gamma ray, increase the effective output area of ​​the detector, and reduce the distortion and artifacts of the reconstructed image.

Description

Technical Field

[0001] The invention relates to a ray positioning method, in particular to an incident ray intelligent positioning method based on a continuous crystal gamma imaging detector. Background Art

[0002] In the process of gamma-ray imaging, the first step is to locate the position of the rays and the scintillator. Therefore, a suitable positioning method can accurately locate the rays and improve some problems caused by system design or hardware. The accuracy of positioning directly affects the performance of the entire imaging system. Improving the ray positioning capability plays an important supporting role in the advancement of gamma cameras.

[0003] At present, the most widely used positioning method in gamma-ray imaging is the classical center of gravity method, also known as the Anger algorithm, which calculates the coordinates of the action point by calculating the center of gravity of the scintillator fluorescence distribution. However, since the center of gravity method is highly sensitive to the changes in the fluorescence distribution inside the crystal during the scintillation event, the algorithm has poor spatial resolution and a small effective imaging area near the edge of the scintillator crystal; in addition, since the center of gravity method is greatly affected by scattering and noise, the positioning of the radiation source based on the center of gravity method is prone to distortion and artifacts. The currently widely used center of gravity positioning method is difficult to meet the performance requirements of gamma-ray imaging equipment in the new era, which seriously restricts the development of gamma-ray imaging equipment. Summary of the invention

[0004] The purpose of the present invention is to provide a method for solving the above-mentioned problems, improving the position linearity of the coded image output by the detector through a neural network, reducing the positioning distortion and distortion caused by the non-uniformity of the detector, and then increasing the effective output area of ​​the detector, so that the coding matrix can contain more effective information, and effectively suppressing artifacts in the process of reconstruction of the radiation source, an incident ray intelligent positioning method based on a continuous crystal gamma imaging detector.

[0005] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows: an incident ray intelligent positioning method based on a continuous crystal gamma imaging detector, wherein the continuous crystal gamma imaging detector comprises a detector and an ADC, and a gamma ray beam is converted into a pulse signal array acting on the incident surface of the detector, and then converted into a response matrix by the ADC, comprising the following steps;

[0006] (1) According to the size of the detector incident surface, a coordinate system is established on the detector incident surface to determine the scanning point, the two-dimensional coordinates of the scanning point, and the scanning path;

[0007] (2) constructing a database, including steps (21) to (22);

[0008] (21) Obtaining a data sample of a scanning point;

[0009] A collimated monoenergetic gamma-ray beam is vertically irradiated on a scanning point, and the continuous crystal gamma imaging detector obtains a response matrix corresponding to several full-energy peak counts of the scanning point within a measurement time t, where N is a positive integer;

[0010] For each response matrix, extract its feature matrix through convolution operation, use the coordinates of the scanning point as the label of the feature matrix, and use the labeled feature matrix as the data sample;

[0011] (22) Obtain data samples of all scanning points according to the scanning path, and construct a database with all data samples;

[0012] (3) KNN positioning; including steps (31)-(34);

[0013] (31) selecting a radiation source at an unknown position, wherein the radiation source releases gamma rays to irradiate the incident surface of the detector, and the continuous crystal gamma imaging detector obtains a response matrix corresponding to several full energy peak counts at the position within a measurement time t;

[0014] (32) extracting the feature matrix of the response matrix obtained in step (31) through a convolution operation, and obtaining several feature matrices corresponding to the response matrix one by one as the feature matrix to be located;

[0015] (33) For a feature matrix to be located, calculate its Euclidean distance with all data samples in the database, find the M data samples with the closest distance, count the labels of the M data samples, and take the coordinates of the scanning point corresponding to the largest number of labels as the position of the feature matrix to be located, where M is a positive integer and less than the total number of data samples;

[0016] (34) The positions of all feature matrices to be located are obtained in turn as the positions of the gamma ray.

[0017] Preferably, in step (1), the coordinate system is established by gridding the incident surface, with each grid center being a scanning point.

[0018] Preferably, the detector adopts a 16×16 SiPM array, the ADC is a 16×16 ADC, and the output is a 16×16 response matrix; the characteristic matrix is ​​8×8.

[0019] Compared with the prior art, the advantages of the present invention are:

[0020] (1) The present invention adopts a machine learning method to locate the detector rays, improves the position linearity of the detector output coded image through the KNN neural network, reduces the positioning distortion and distortion caused by the non-uniformity of the detector, thereby increasing the effective response area of ​​the detector and reducing the R&D cost; compared with the traditional positioning method, this intelligent positioning system overcomes the influence of the compression effect, so it can be applied to thick crystal detectors, effectively reducing the measurement time required to locate the radiation source.

[0021] (2) The present invention improves the detector positioning algorithm, improves the accuracy of the positioning algorithm for ray positioning, increases the effective output area of ​​the detector, and makes the encoding matrix contain more effective information, thereby improving the quality of radiation source reconstruction;

[0022] (3) The intelligent positioning method proposed in the present invention corrects the nonlinear parameters in the fitting process through a neural network, eliminating the non-uniformity of the detector and the changes in the measurement environment. Therefore, it has strong robustness and improves the anti-interference ability of the instrument. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of radiation irradiating a continuous crystal gamma imaging detector;

[0024] Figure 2 It is a flow chart of the present invention;

[0025] Figure 3 A schematic diagram of a scanning path in the present invention;

[0026] Figure 4 Schematic diagram of gamma-ray beam irradiating the detector surface;

[0027] Figure 5 A comparison diagram of the pan-field ray positioning obtained by the traditional method and the present invention;

[0028] Figure 6 A comparison diagram of gamma rays emitted by a radioactive source after being positioned by a conventional method and the method of the present invention, and after the radioactive source is imaged by a gamma camera;

[0029] Figure 7 This is a comparison diagram of gamma rays emitted by another radioactive source after being positioned by a traditional method and the method of the present invention, and the radioactive source being imaged by a gamma camera.

[0030] In the figure: 1. Collimator unit; 2. Detector. DETAILED DESCRIPTION

[0031] The present invention will be further described below in conjunction with the accompanying drawings.

[0032] Example 1: See Figure 1-Figure 3, an incident ray intelligent positioning method based on a continuous crystal gamma imaging detector, the continuous crystal gamma imaging detector includes a detector 2 and an ADC, the gamma ray beam acts on the incident surface of the detector 2 and is converted into a pulse signal array, and then converted into a response matrix by the ADC, the positioning method includes the following steps;

[0033] (1) According to the size of the incident surface of detector 2, a coordinate system is established on the incident surface of detector 2 to determine the scanning point, the two-dimensional coordinates of the scanning point, and the scanning path;

[0034] (2) constructing a database, including steps (21) to (22);

[0035] (21) Obtaining a data sample of a scanning point;

[0036] A collimated monoenergetic gamma-ray beam is vertically irradiated on a scanning point, and the continuous crystal gamma imaging detector obtains a response matrix corresponding to several full-energy peak counts of the scanning point within a measurement time t, where N is a positive integer;

[0037] For each response matrix, extract its feature matrix through convolution operation, use the coordinates of the scanning point as the label of the feature matrix, and use the labeled feature matrix as the data sample;

[0038] (22) Obtain data samples of all scanning points according to the scanning path, and construct a database with all data samples;

[0039] (3) KNN positioning; including steps (31)-(34);

[0040] (31) A radiation source at an unknown position is selected, the radiation source releases gamma rays to irradiate the incident surface of the detector 2, and the continuous crystal gamma imaging detector obtains a response matrix corresponding to several full energy peak counts at the position within a measurement time t;

[0041] (32) extracting the feature matrix of the response matrix obtained in step (31) through a convolution operation, and obtaining several feature matrices corresponding to the response matrix one by one as the feature matrix to be located;

[0042] (33) For a feature matrix to be located, calculate its Euclidean distance with all data samples in the database, find the M data samples with the closest distance, count the labels of the M data samples, and take the coordinates of the scanning point corresponding to the largest number of labels as the position of the feature matrix to be located, where M is a positive integer and less than the total number of data samples;

[0043] (34) The positions of all feature matrices to be located are obtained in turn as the positions of the gamma ray.

[0044] In addition, in step (1), the coordinate system is established by gridding the incident surface, with each grid center being a scanning point. The detector 2 uses a 16×16 SiPM array, so the ADC is a 16×16 ADC, and the output is a 16×16 response matrix; the characteristic matrix is ​​8×8.

[0045] Example 2: See Figure 1-Figure 7 We further define the method based on Example 1. An incident ray intelligent positioning method based on a continuous crystal gamma imaging detector, wherein the continuous crystal gamma imaging detector includes a detector 2 and an ADC, and a gamma ray beam is converted into a pulse signal array when acting on the incident surface of the detector 2, and then converted into a response matrix by the ADC.

[0046] Among them, the size of detector 2 is as follows Figure 3 ,The size of detector 2 is 100mm×100mm, and the number of scanning points is determined by taking 1mm as a scanning interval, with a total of 100×100=10000 scanning points. Based on these scanning points, we can obtain the spatial coordinates of each scanning point and plan the scanning route;

[0047] Detector 2 uses a 16×16 SiPM array to output a 16×16 pulse signal array. The ADC uses a 256-channel high-speed ADC to convert the 16×16 pulse signal array into a 16×16 response matrix.

[0048] The convolution operation is for feature extraction. The feature matrix obtained by the convolution operation can be 10×10 or 8×8, etc.

[0049] Regarding the collimated monoenergetic gamma ray beam, in this embodiment, the monoenergetic gamma ray beam is converted into a collimated monoenergetic gamma ray beam by the collimator unit 1. For details, see Figure 4 The collimator unit 1 is divided into two parts: a tungsten collimator and a lead collimator. The two collimators are two coaxially arranged cylinders with an opening of 1 mm in diameter in the middle; the radius of the tungsten collimator is 20 mm and the height is 20 mm, the radius of the lead collimator is 30 mm and the height is 50 mm, and the lead collimator is closer to the incident surface of the detector 2; the tungsten collimator is used to place the 137Cs radiation source to reduce the radiation dose; the lead collimator is used to collimate gamma rays, and the emission direction of the gamma rays after passing through the collimator unit 1 is perpendicular to the crystal incident surface of the detector 2.

[0050] In step (31), the continuous crystal gamma imaging detector obtains the number of response matrices corresponding to the full energy peak count at the position within the measurement time t, which is related to the activity of the radiation source. In this embodiment, within the measurement time t, a total of 150 response matrices corresponding to the full energy peak count are obtained;

[0051] In step (33), M is a positive integer and is less than the total number of data samples. In this embodiment, M=100. Then, for a feature matrix to be located, there are 100 data samples that are closest to each other. We count the labels of these 100 data samples and take the coordinates of the scanning point corresponding to the label with the largest number as the position of the feature matrix to be located.

[0052] In order to illustrate the technical effect of the present invention, we compare the pan-field ray positioning diagram obtained by the traditional method and the present invention, see Figure 5 ,from Figure 5 It can be seen that in the traditional method, there is a large detection dead zone, such as Figure 5 The black frame area outside the imaging area on the left side has almost no dead zone, so the effective imaging area of ​​the present invention is much larger than the traditional positioning algorithm, the effective area is close to the incident surface area of ​​the detector 2, the edge compression effect of the detector 2 is alleviated, and the unevenness is improved.

[0053] In addition, we used two different radioactive sources to emit gamma rays, and positioned them using the traditional method and the method of the present invention respectively, and then imaged the radioactive sources using a gamma camera, and compared the images to obtain Figure 6 and Figure 7 .from Figure 6 and Figure 7 It can be seen from the figure that the image signal artifacts of the imaging of the present invention are reduced, the signal-to-noise ratio is improved, the image edges are clear, and the image quality is significantly enhanced.

[0054] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An incident ray intelligent positioning method based on a continuous crystal gamma imaging detector, wherein the continuous crystal gamma imaging detector comprises a detector and an ADC, wherein a gamma ray beam is converted into a pulse signal array when acting on the incident surface of the detector, and then converted into a response matrix by the ADC, characterized in that: The steps include: (1) According to the size of the detector incident surface, a coordinate system is established on the detector incident surface to determine the scanning point, the two-dimensional coordinates of the scanning point, and the scanning path; (2) constructing a database, including steps (21) to (22); (21) Obtaining a data sample of a scanning point; A collimated monoenergetic gamma-ray beam is vertically irradiated on a scanning point, and the continuous crystal gamma imaging detector obtains a response matrix corresponding to several full-energy peak counts of the scanning point within a measurement time t, where N is a positive integer; For each response matrix, extract its feature matrix through convolution operation, use the coordinates of the scanning point as the label of the feature matrix, and use the labeled feature matrix as the data sample; (22) Obtain data samples of all scanning points according to the scanning path, and construct a database with all data samples; (3) KNN positioning; including steps (31)-(34); (31) selecting a radiation source at an unknown position, wherein the radiation source releases gamma rays to irradiate the incident surface of the detector, and the continuous crystal gamma imaging detector obtains a response matrix corresponding to several full energy peak counts at the position within a measurement time t; (32) extracting the feature matrix of the response matrix obtained in step (31) through a convolution operation, and obtaining several feature matrices corresponding to the response matrix one by one as the feature matrix to be located; (33) For a feature matrix to be located, calculate its Euclidean distance with all data samples in the database, find the M data samples with the closest distance, count the labels of the M data samples, and take the coordinates of the scanning point corresponding to the largest number of labels as the position of the feature matrix to be located, where M is a positive integer and less than the total number of data samples; (34) The positions of all feature matrices to be located are obtained in turn as the positions of the gamma ray.

2. The incident ray intelligent positioning method based on the continuous crystal gamma imaging detector according to claim 1 is characterized in that: In step (1), the coordinate system is established by meshing the incident surface, with each mesh center being a scanning point.

3. The incident ray intelligent positioning method based on the continuous crystal gamma imaging detector according to claim 1 is characterized in that: The detector adopts a 16×16 SiPM array, so the ADC is a 16×16 ADC, and the output is a 16×16 response matrix; the characteristic matrix is ​​8×8.

Images