Fast 3D Imaging Method and System for Compton Cameras in the Far Field

Abstract

This invention relates to a rapid 3D imaging method and system using a Compton camera for far-field global imaging, and pertains to the field of nuclear radiation detection and application technology. According to the method of this invention, firstly, the solid angle distribution of a radioactive source is acquired using a Compton camera at two spatial viewpoints; then, starting from the detector center position of the camera, two sets of 3D spatial vectors are constructed based on the solid angle distribution; finally, the 3D position distribution of the radioactive source is obtained by intersecting the two sets of vectors. This invention utilizes the intersection of two sets of vectors, calculating only the distribution and probability of the intersection points, thus obtaining the 3D position distribution of the radioactive source without traversing the entire imaging space, thereby saving computer memory and computation time, and significantly improving the 3D imaging speed of the Compton camera.

Description

Technical Field

[0001] This invention relates to the field of nuclear radiation detection and application technology, and in particular to a method and system for rapid three-dimensional imaging with a Compton camera for far-field global coverage. Background Technology

[0002] Compton imaging (Compton camera) is a gamma-ray imaging technique based on the Compton scattering effect, which can reconstruct the spatial distribution image of a radiation source without the need for a mechanical collimator. Its imaging principle relies on Compton scattering of photons with matter. The scattering angle is determined by the equations of Compton scattering. By detecting the energy and scattering direction of the scattered photons and recoil electrons, the direction of the radiation source can be determined on a Compton cone with the opposite direction of the scattered photon emission as its axis and the Compton scattering angle as its half-apex angle. Each scattering event forms a cone, and the intersection of these cones theoretically indicates the spatial location of the radiation source.

[0003] Image reconstruction algorithms for the Compton camera mainly include analytical reconstruction algorithms and iterative reconstruction algorithms. Analytical reconstruction algorithms directly backproject the detector's recorded response onto all possible spatial locations, also known as backprojection imaging. As detector data accumulates, the true location of the radiation source is gradually enhanced and highlighted. However, because many background voxels are also assigned probabilities during reconstruction, images generated by analytical algorithms typically have poor resolution. Iterative reconstruction methods usually solve for the probability distribution of the radiation source in the imaging space based on the transfer relationship between the image space and the detector response. Each iteration corrects the image estimate, and multiple iterations continue until convergence. Compared to analytical algorithms, iterative methods can more accurately handle complex image structures and background noise, typically yielding higher-quality images.

[0004] In near-field imaging (where the distance between the radiation source and the detector is only a few centimeters), the three-dimensional information of the radiation source's location can be directly obtained by superimposing each Compton cone, and its true location is significantly enhanced by the contribution of each cone. However, when the radiation source is far from the detector, the differences in the distribution of the Compton cone vertices within the detector are difficult to reflect, resulting in the entire area in the direction of the radiation source being contributed by the Compton cones, making it impossible to accurately obtain the distance information of the radiation source. Therefore, it is necessary to collect Compton scattering event data from multiple spatial locations, so that the location of the radiation source is contributed by Compton cones from different angles, thereby correctly reconstructing the three-dimensional image of the radiation source.

[0005] Traditional 3D Compton imaging requires dividing the imaging space into a 3D mesh to reconstruct the 3D distribution of the radiation source within the mesh. To obtain the location information of radiation sources at greater distances from the detector, the required imaging space area increases significantly. As the imaging space and the number of meshes increase, the computational load rises dramatically, making it difficult to meet real-time imaging requirements. Summary of the Invention

[0006] One of the objectives of this invention is to provide a method for achieving rapid 3D imaging with a Compton camera in the far-field global domain, thereby improving the imaging speed of the Compton camera in practical application scenarios.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A fast 3D imaging method for Compton cameras with far-field global coverage includes the following steps:

[0009] (1) Solid angle distribution reconstruction:

[0010] By using data acquired by a Compton camera at two different spatial locations, the solid angle of the radiation source was reconstructed, thus obtaining the solid angle of the same radiation source from the perspectives of the two spatial locations. Distribution and its corresponding probability;

[0011] (2) Vector construction:

[0012] Starting from the detector center position of two cameras at different spatial locations, and taking the solid angle of the radiation source from the perspective of each of the two spatial locations... Using distribution as direction, two sets of three-dimensional spatial vectors are constructed;

[0013] (3) Calculation of vector intersection points:

[0014] Intersect two sets of three-dimensional space vectors, calculate the location of the intersection point, and calculate the probability of the intersection point based on the probability of the vector directions;

[0015] (4) Three-dimensional distribution construction:

[0016] Based on the calculated intersection points and probabilities of all vectors, a three-dimensional location distribution of the radioactive source is constructed.

[0017] Furthermore, the following steps are included before step (1):

[0018] The Compton camera was used to detect radiation sources at two different locations, collect data over a certain period of time, process the data, filter out Compton imaging events, and then reconstruct the solid angle distribution of the radiation sources.

[0019] Further, in step (1), the solid angle distribution of the radioactive source is reconstructed according to the following steps:

[0020] First, the entire imaging space is scanned quickly using a large angular step size, and the focal region where the radioactive source may exist is quickly determined by reconstructing the detected Compton's cone analytically.

[0021] Then, these focal regions are finely divided with small angular steps, and the solid angle distribution of the radiation source is reconstructed by combining iterative algorithms, thereby quickly obtaining a high-precision solid angle distribution of the radiation source.

[0022] Furthermore, the ratio between the larger angle step size and the smaller angle step size is 10:1. For example, the larger angle step size is 10°, and the smaller angle step size is 1°.

[0023] Furthermore, in step (3), during the calculation of vector intersection points, for vector pairs with different starting points and not directly intersecting, the point closest to them in space is taken as the intersection point, and this intersection point is calculated.

[0024] Furthermore, in step (3), the nearest point of a pair of vectors with different starting points and that do not intersect directly is calculated using the following formula, and this point is taken as the intersection point P of the vector pair. ij :

[0025]

[0026] Where P1 and P2 are the coordinates of the Compton camera at two different locations, d i d j To extract the direction from the solid angle distribution, P ij Starting from P1 and P2, d i d j The intersection of two vectors constructed with respect to the direction.

[0027] Further, in step (4), the probabilities of all vector intersections are calculated through the following steps:

[0028] Extracting d from the solid angle distribution i and d j The probability of each vector intersection point P is given. ij The corresponding d in the middle i and d j The probabilities are multiplied together to obtain the intersection point P of each vector. ij The probability of.

[0029] Furthermore, in step (1), a spherical coordinate system is used to divide the imaging space in order to reconstruct the solid angle distribution of the radiation source, with the origin of the coordinate system at each location located at the center of the detector in the camera.

[0030] Another object of the present invention is to provide a fast 3D imaging system for a Compton camera oriented towards the far-field global field, which includes a storage unit and a computing unit. The storage unit stores a fast 3D imaging program for a Compton camera oriented towards the far-field global field, and the fast 3D imaging program for a Compton camera oriented towards the far-field global field is run by the computing unit to perform the steps in the aforementioned fast 3D imaging method for a Compton camera oriented towards the far-field global field.

[0031] Traditional 3D Compton imaging relies on constructing a 3D mesh within the imaging space and reconstructing the 3D distribution of the radiation source using this mesh. This method often encounters the problem of excessively long imaging times when dealing with far-field radiation sources due to the need to process the large 3D mesh. In contrast, this invention proposes a novel vector intersection method that achieves 3D imaging using 2D imaging technology, effectively solving the problem of excessively long imaging times caused by the large 3D mesh. This method avoids processing a large number of 3D meshes by reconstructing the solid angle distribution from data acquired by the Compton camera at two different locations, and then using the solid angle distribution from the two viewpoints... By constructing vectors that intersect pairwise and combining them with corresponding probability coefficients, the three-dimensional position distribution of the radiation source can be directly constructed, thereby significantly improving imaging efficiency and real-time performance.

[0032] The core idea of ​​this invention is that the location of a radiation source can be found simply by calculating the intersection of two sets of vectors, thus avoiding the need for traversing the entire three-dimensional spatial grid in traditional three-dimensional Compton imaging and saving computation time. Specifically, vectors are extracted from two detection positions and their corresponding solid angle distributions, and the intersections of the two sets of vectors are calculated. These intersections represent the locations where radiation sources may exist in three-dimensional space. By multiplying the probabilities of each vector pair in the solid angle distribution, the probability of the vector pair intersection is obtained, thereby constructing the three-dimensional location distribution of the radiation source. This method significantly reduces the number of grid cells traversed, reduces the computational workload of searching and verifying Compton cone information for grid cells, and thus greatly reduces computation time and memory consumption, improving imaging speed. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the vector intersection method in the embodiment;

[0034] Figure 2 This is a schematic diagram of the imaging space division in the embodiment;

[0035] Figure 3 This is a flowchart illustrating the fast 3D imaging method using a Compton camera for far-field global imaging in this embodiment.

[0036] Figure 4The images shown in the embodiments include solid angle distribution images measured twice and a final three-dimensional position distribution image; wherein, (a) is a solid angle distribution image of the Compton camera in the first position, (b) is a solid angle distribution image of the Compton camera in the second position, and (c) is a three-dimensional position distribution image of the radiation source. Detailed Implementation

[0037] To help those skilled in the art better understand the improvements of this invention compared to the prior art, the invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0038] The objective of this invention is to provide a method for achieving rapid 3D imaging with a Compton camera across the entire far-field domain, thereby improving the imaging efficiency of Compton cameras in practical applications. Addressing the challenges faced by traditional 3D Compton imaging techniques in far-field radiation source measurement, particularly the excessively long imaging time caused by handling large-scale 3D meshes, this invention proposes a novel vector intersection method. This method achieves 3D imaging through vector intersection, effectively solving the bottleneck problems of traditional methods.

[0039] The core idea of ​​this invention is that the Compton camera extracts vectors from two detection locations and their corresponding solid angle distributions, and calculates their intersection to determine the location of the radiation source. This method avoids the complex process of searching and calculating Compton cone information for each three-dimensional mesh cell, thus significantly saving computation time and memory consumption.

[0040] Specifically, the radiation source direction information provided by the Compton camera at two different locations, due to their different starting points, can construct intersecting vector pairs. The intersection point of these vector pairs is the location of the radiation source. Considering measurement errors, the actual vector pairs may not intersect precisely. Therefore, the point in space closest to these two vectors can be considered the intersection point, and this intersection point can be calculated mathematically. Combining the probability of each vector's direction in the solid angle distribution, the final distribution of vector intersection points is the three-dimensional location distribution of the radiation source.

[0041] To optimize computational resource allocation and reduce background computational consumption, this invention employs a double-division imaging space approach in solid angle distribution reconstruction. First, the imaging space is divided with a large angular step size, and an analytical reconstruction algorithm is used to quickly scan the entire space to determine the "focal" region where the radiation source is located. Then, within these "focal" regions, an iterative algorithm is combined for high-precision reconstruction, significantly improving both imaging speed and accuracy.

[0042] like Figure 1As shown, this invention proposes a fast 3D imaging method based on vector intersection, effectively resolving the contradiction between time efficiency and imaging spatial range in far-field 3D imaging. This method utilizes data acquired by a Compton camera at two different locations to reconstruct the solid angle distribution. Then, vectors containing probabilistic information are extracted from the solid angle distribution, and the 3D localization and imaging of the radiation source are achieved by calculating the intersection points of these vectors. Compared with traditional methods, this method reduces the spatial dimension in Compton imaging, improves imaging speed, and can cover the entire far-field imaging range.

[0043] The following will describe in detail the principle and structure of the Compton camera 3D fast Compton imaging method for far-field global coverage described in this invention, with reference to embodiments. The general flow of this method can be found in [reference needed]. Figure 3 .

[0044] This embodiment uses Geant4 to conduct Monte Carlo simulation tests to verify the feasibility of the solution. The model used is a 44×44×10mm model. 3 The CZT detector is internally divided into an array of 484 pixels, each measuring 2×2×10mm. 3 In the simulation, two CZT detectors were constructed at (-3,0,0)m and (3,0,0)m respectively, and a γ-ray point source with an energy of 662keV was set at (0,0,4)m.

[0045] The implementation steps are summarized as follows:

[0046] First, the radiation source was detected at two preset locations (coordinates (-3,0,0) m and (3,0,0) m). During data processing, the detector was assigned an energy resolution of 1.5% and a depth-direction position resolution of 2 mm. Subsequently, Compton imaging events were selected from the raw data. Through this process, approximately 2500 Compton imaging events were obtained from the two measurement locations.

[0047] Next, two-dimensional imaging of the radiation source is performed. For example... Figure 2 As shown, in the two-dimensional reconstruction process, a spherical coordinate system is used to divide the imaging space to reconstruct the solid angle distribution image of the radiation source. Simultaneously, by doubly dividing the imaging space, the computational load on the background space is reduced. The origin of the two-dimensional reconstruction coordinate system at each location is set at the center of the corresponding detector, ensuring that the imaging range covers the entire 4π solid angle space. In the spherical coordinate system, θ represents the angle between a ray in a certain direction in space and the Z-axis, and its value ranges from 0 to π. It represents the angle formed by the counterclockwise rotation between the projection of the directional ray onto the XY plane and the positive X-axis, and its value ranges from 0 to 2π.

[0048] To optimize the allocation of computational resources and reduce the computational cost of the background angle, the imaging space was first divided with a large angular step size (10°), and the entire space was quickly scanned. During this process, several "focal" regions where the radioactive source might exist were preliminarily determined by analytical reconstruction of the detected Compton's cone.

[0049] Subsequently, within these "focal" regions, fine division is performed with a smaller angular step size (1°), and the solid angle distribution of the radioactive source is reconstructed using an iterative algorithm, thereby quickly obtaining a high-precision solid angle distribution of the radioactive source.

[0050] This embodiment uses two-dimensional reconstruction instead of three-dimensional reconstruction and double-divides the imaging space, effectively reducing the number of grids during reconstruction and significantly lowering memory and computation time consumption. This embodiment performs high-precision calculations only within the "focal" region, optimizing the allocation of computational resources and effectively avoiding the redundant burden of high-resolution calculations across the entire reconstruction space.

[0051] The following is for reference Figure 4 The experimental results illustrate the vector intersection imaging process. The detector measures the radiation source at two positions, P1 and P2. Each measurement provides information on the solid angle distribution of the radiation source, thus generating an image with starting points P1 and P2 and direction d. i d j The vectors are denoted by . By arranging and combining two sets of vectors, and using the following formula to calculate the point in space closest to the two vectors, this point is considered the intersection point P of the vectors. ij :

[0052]

[0053] P ij The probability is d i and d j The product of probabilities, calculated by plotting all vector intersections P. ij The distribution of the radiation source and its corresponding probability were normalized to generate a three-dimensional distribution image. The location result in the test was (10.7, 7.7, 387.4) cm, which is close to the actual location of the radiation source.

[0054] Test results show that the vector intersection method can generate a 3D image of the radiation source distribution in just 35 seconds, compared to 1485 seconds for the traditional 3D mesh imaging method using the Compton conic intersection method, a difference of more than 42 times. This is mainly because the Compton conic intersection method divides the 3D space into 216,000 grids, and the system matrix occupies as much as 8.64GB of memory during iterative reconstruction, placing a significant burden on the computer. In contrast, the vector intersection method only divides the space into 3D grids for calculation, resulting in approximately 55 times fewer elements in the imaging space, thus achieving a speed improvement of more than 42 times in 3D imaging. Furthermore, the imaging range corresponding to the 3D space divided by the Compton conic intersection method in the experiment was 6×6×6m. 3 The vector intersection method, in theory, can perform three-dimensional imaging of radioactive sources at any location.

[0055] In summary, this embodiment provides a fast 3D imaging method for a Compton camera covering the entire far-field domain. By simplifying the computational process, it only needs to process the solid angle distribution, eliminating the need to calculate the distribution of the Compton cone in each 3D grid, thus significantly reducing computational complexity. Furthermore, this method does not require a pre-defined fixed imaging space and can directly generate the 3D distribution of the radiation source from the calculation results, resulting in a larger solid angle range for imaging. This method not only significantly shortens the far-field 3D imaging time but also provides a far-field global imaging range, demonstrating promising application prospects.

[0056] In addition, the present invention can be implemented in other ways, and any obvious substitutions without departing from the concept of the present invention are within the protection scope of the present invention.

[0057] To facilitate understanding by those skilled in the art of the improvements of this invention over the prior art, some of the accompanying drawings and descriptions have been simplified, and for clarity, some other elements have been omitted from this application. Those skilled in the art should realize that these omitted elements may also constitute the content of this invention.

Claims

1. A fast 3D imaging method for Compton cameras with far-field global coverage, characterized in that, Includes the following steps: (1) Solid angle distribution reconstruction: By using data acquired by a Compton camera at two different spatial locations, the solid angle of the radiation source was reconstructed, thus obtaining the solid angle of the same radiation source from the perspectives of the two spatial locations. Distribution and its corresponding probability; (2) Vector construction: Starting from the detector center position of two cameras at different spatial locations, and taking the solid angle of the radiation source from the perspective of each of the two spatial locations... Using distribution as direction, two sets of three-dimensional spatial vectors are constructed; (3) Calculation of vector intersection points: Intersect two sets of three-dimensional space vectors, calculate the location of the intersection point, and calculate the probability of the intersection point based on the probability of the vector directions; (4) Three-dimensional distribution construction: Based on the calculated intersection points and probabilities of all vectors, a three-dimensional location distribution of the radioactive source is constructed.

2. The fast 3D imaging method for Compton cameras oriented towards far-field global coverage as described in claim 1, characterized in that: In step (1), the solid angle distribution of the radioactive source is reconstructed according to the following steps: First, the entire imaging space is scanned quickly using a large angular step size, and the detected Compton event is analyzed and reconstructed to quickly determine the focal region where the radiation source may exist. Then, these focal regions are finely divided with small angular steps, and the solid angle distribution of the radiation source is reconstructed by combining iterative algorithms, so as to quickly obtain a high-precision solid angle distribution of the radiation source.

3. The fast 3D imaging method for Compton cameras oriented towards far-field global coverage according to claim 2, characterized in that, In step (3), during the calculation of vector intersection points, for vector pairs with different starting points and not directly intersecting, the point closest to them in space is taken as the intersection point, and this intersection point is calculated.

4. The fast 3D imaging method for Compton cameras oriented towards far-field global coverage as described in claim 3, characterized in that: The nearest point of a pair of vectors in space that have different starting points and do not intersect directly is calculated using the following formula, and this point is taken as the intersection point P of the vector pair. ij : Where P1 and P2 are the coordinates of the Compton camera at two different locations, d i d j To extract the direction from the solid angle distribution, P ij Starting from P1 and P2, d i d j The intersection of two vectors constructed with respect to the direction.

5. The fast 3D imaging method for Compton cameras oriented towards far-field global coverage according to claim 4, characterized in that: In step (4), the probabilities of all vector intersections are calculated using the following steps: Extracting d from the solid angle distribution i and d j The probability of each vector intersection point P is given. ij The corresponding d in the middle i and d j The probabilities are multiplied together to obtain the intersection points P of each vector. ij The probability of.

6. A Compton camera fast 3D imaging system for far-field global coverage, comprising a storage unit and a computing unit, wherein the storage unit stores a Compton camera fast 3D imaging program for far-field global coverage, and the Compton camera fast 3D imaging program for far-field global coverage is run by the computing unit to execute the steps of the Compton camera fast 3D imaging method for far-field global coverage according to any one of claims 1-5.

Images