Static CT acquisition system

Abstract

The invention relates to a static CT acquisition system, belongs to the technical field of object reconstruction, and solves the problems of low object image acquisition efficiency and low reconstructed image quality in the prior art. The static CT acquisition system comprises at least one static acquisition device which is sequentially arranged according to an object scanning sequence, and each static acquisition device is the same in structure and comprises a base, a radiation source array and a detector array; the radiation source array is arranged in the base, radiation source windows are formed in the inner side of the base every other same preset area, and each radiation source window corresponds to one radiation source; the detector array comprises a first detector sub-array, a second detector sub-array and a third detector sub-array; a first detector sub-array is arranged in each preset area on the inner side of the base, and a second detector sub-array and a third detector sub-array are arranged in other areas, except the preset areas and the radiation source window, on the inner side of the base. And the reconstruction precision and efficiency of the object are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of object reconstruction, and in particular to a static CT acquisition system. Background Art

[0002] At present, X-ray machines, as a non-invasive inspection equipment, have been widely used in various security inspection scenarios, such as airports, ports, hospitals and scenic spots.

[0003] Traditional security CT systems use a spiral CT solution, rotating the X-ray source and detectors, utilizing key components such as bearings and slip rings to achieve 360° projection data acquisition and complete CT imaging. While increasing the rotation speed and the number of detector rows can improve detection efficiency, the former is limited by the centrifugal acceleration of the mechanical system, while the latter can affect image quality due to the large imaging cone angle. Furthermore, when reconstructing objects from images captured by traditional security CT systems, the resulting object reconstruction accuracy is low, failing to meet the demands of increasingly sophisticated imaging.

[0004] Therefore, there is an urgent need for a new technical solution for collecting and reconstructing objects. Summary of the Invention

[0005] In view of the above analysis, an embodiment of the present invention aims to provide a static CT acquisition system to solve the problems of low object image acquisition efficiency and low reconstructed image quality in the prior art.

[0006] An embodiment of the present invention provides a static CT acquisition system, comprising at least one static acquisition device arranged in sequence according to an object scanning order, each static acquisition device having the same structure, including a base, a radiation source array, and a detector array;

[0007] The ray source array is arranged inside the base, and ray source windows are opened on the inner side of the base at every same preset area, and each ray source window corresponds to a ray source; wherein the distribution angle of the ray source array is greater than or equal to π+2γ and less than 2π, where γ represents the opening angle of the fan beam of the ray source;

[0008] The detector array includes a first detector sub-array, a second detector sub-array and a third detector sub-array; the first detector sub-array is arranged at each preset area inside the base, and the second detector sub-array and the third detector sub-array are arranged in other areas inside the base except the preset areas and the radiation source window.

[0009] Based on further improvement of the above system, the shape of the base is any of the following:

[0010] Ring;

[0011] square;

[0012] Regular hexagon.

[0013] Based on a further improvement of the above system, the first detector subarray, the second detector subarray and the third detector subarray are respectively connected to the base via fixing bolts.

[0014] Based on the further improvement of the above system, the static CT acquisition system further includes a ray source window controller;

[0015] The ray source window controller is used to control the working time of each ray source in all ray source arrays;

[0016] The detector array receives the ray source data and transmits the received ray source data as a detection projection image to the ray source window controller;

[0017] The ray source window controller is also used to reconstruct a target reconstructed image based on the detected projection image.

[0018] Based on a further improvement of the above system, the target reconstructed image is obtained by reconstructing the detected projection image, including:

[0019] Based on the linear interpolation method, the detected projection image is interpolated in the projection viewing angle direction to obtain the interpolated restored projection image; the interpolated restored projection image is input into the pre-trained projection domain network model to obtain the optimized projection image;

[0020] The optimized projection image and the detection projection image are reconstructed based on a preset image reconstruction algorithm to obtain an optimized reconstructed image and an initial reconstructed image;

[0021] The optimized reconstructed image and the initial reconstructed image are input into the pre-trained image domain network model to obtain the target reconstructed image.

[0022] Based on the further improvement of the above system, the image domain network model is a dual encoder-single decoder network, which includes a first feature extraction module and a second feature extraction module with the same structure, as well as a feature fusion module;

[0023] The first feature extraction module and the second feature extraction module are used to perform multi-scale feature extraction on the optimized reconstructed image and the initial reconstructed image, respectively, to obtain a multi-scale optimized feature map and a multi-scale initial feature map;

[0024] The feature fusion module is used to splice and fuse the optimized feature map and the initial feature map of the same scale to obtain the target reconstructed image;

[0025] The first feature extraction module and the second feature extraction module have the same structure, and both include a plurality of dual convolution modules connected in sequence; the dual convolution module includes two convolution modules connected in sequence, each convolution module is used to perform a convolution operation, a normalization operation and a ReLU activation function operation on the input feature map;

[0026] The first dual convolution module is used to receive the optimized reconstructed image or the initial reconstructed image, and transmit the obtained feature map to the next dual convolution module and output it to the feature fusion module; the middle dual convolution module is used to obtain the feature map of the current dual convolution module based on the feature map output by the previous dual convolution module, and transmit the feature map of the current dual convolution module to the next dual convolution module and output it to the feature fusion module; the last dual convolution module is used to obtain the feature map of the last dual convolution module based on the feature map output by the previous dual convolution module, and output the feature map of the last dual convolution module to the feature fusion module.

[0027] Based on the further improvement of the above system, the feature fusion module includes a spliced ​​two-dimensional deconvolution module, multiple spliced ​​convolution and two-dimensional deconvolution modules and a spliced ​​double convolution splicing module connected in sequence;

[0028] A splicing two-dimensional deconvolution module is used to receive the feature map of the last double convolution module in the first feature extraction module and the second feature extraction module, and perform splicing and two-dimensional deconvolution operations, and output the obtained feature map to the first splicing convolution and two-dimensional deconvolution module; each splicing convolution and two-dimensional deconvolution module is used to perform splicing, convolution, normalization, ReLU activation function and two-dimensional deconvolution on the input feature map;

[0029] The first splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the splicing two-dimensional deconvolution module, as well as the feature maps output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the splicing two-dimensional deconvolution module, and transmit the obtained feature map to the next splicing convolution and two-dimensional deconvolution module;

[0030] The intermediate splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the previous splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the previous splicing convolution and two-dimensional deconvolution module, and transmit the obtained feature map to the next splicing convolution and two-dimensional deconvolution module;

[0031] The last splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the previous splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the previous splicing convolution and two-dimensional deconvolution module, and transmit the obtained feature map to the splicing double convolution splicing module;

[0032] The splicing double convolution splicing module is used to receive the feature map output by the last splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the last splicing convolution and two-dimensional deconvolution module, and perform splicing operation, convolution operation, normalization operation, ReLU activation function operation, convolution operation, normalization operation and ReLU activation function operation on the input feature map in sequence, and perform splicing operation on the obtained feature map and the optimized reconstructed image as the target reconstructed image.

[0033] Based on the further improvement of the above system, the projection domain network model includes a feature extraction module, an upsampling module and a splicing module connected in sequence;

[0034] The feature extraction module is used to extract multi-scale features from the input interpolated and restored projection image, and input the obtained feature maps of multiple scales into the upsampling module:

[0035] An upsampling module is used to perform layer-by-layer upsampling and splicing of multiple feature maps of different scales to obtain a tenth feature map, and output the obtained tenth feature map to the splicing module;

[0036] The splicing module is used to splice the input interpolated restored projection image and the tenth feature map, and use the spliced ​​feature map as the optimized projection image.

[0037] Based on a further improvement of the above system, the feature extraction module includes a first deformable convolution module, a preset number of first dual multi-scale deformable convolution modules and a second dual multi-scale deformable convolution module connected in sequence;

[0038] A first deformable convolution module is used to perform a deformable convolution operation and a multi-scale deformable convolution operation on the input interpolated restored projection image to obtain a first feature map;

[0039] The first first dual multi-scale deformable convolution module performs a dual multi-scale deformable convolution operation on the first feature map to obtain a second feature map; multiple intermediate first dual multi-scale deformable convolution modules respectively perform a dual multi-scale deformable convolution operation on the feature map output by the previous first dual multi-scale deformable convolution module to obtain feature maps of different scales as multiple third feature maps; the last first dual multi-scale deformable convolution module performs a dual multi-scale deformable convolution operation on the feature map output by the previous first dual multi-scale deformable convolution module to obtain a fourth feature map;

[0040] The second dual multi-scale deformable convolution module is used to perform a dual multi-scale deformable convolution operation on the fourth feature map to obtain a fifth feature map.

[0041] Based on the further improvement of the above system, the upsampling module includes a two-dimensional deconvolution module, a preset number of deformable convolution and deconvolution modules and a second deformable convolution module;

[0042] The two-dimensional deconvolution module performs a two-dimensional deconvolution operation on the input fifth feature map to obtain a sixth feature map;

[0043] The sixth feature map is concatenated with the fourth feature map of the same scale and input into the first deformable convolution and deconvolution module. Each deformable convolution and deconvolution module is used to perform a deformable convolution operation and a two-dimensional deconvolution operation on the input feature map, and output feature maps of different scales.

[0044] The input of each deformable convolution and deconvolution module is the feature map obtained by concatenating the feature map output by the previous deformable convolution and deconvolution module and the third feature map of the same scale output by the feature extraction module;

[0045] The input of the last deformable convolution and deconvolution module is the concatenation of the feature map output by the previous deformable convolution and deconvolution module and the second feature map of the same scale;

[0046] The feature map output by the last deformable convolution and deconvolution module is concatenated with the first feature map of the same scale and input into the second deformable convolution module to obtain the tenth feature map.

[0047] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0048] 1. By arranging the radiation source array inside the base, opening radiation source windows inside the base at every identical preset area, and deploying a first detector sub-array at each preset area inside the base, and deploying a second detector sub-array and a third detector sub-array in areas other than the preset areas and radiation source windows inside the base, the operating time of each radiation source is controlled during use, thereby reducing the manufacturing difficulty and cost of the CT acquisition equipment, eliminating the need for additional movement of the entire static acquisition device during operation, and improving the acquisition efficiency of the object;

[0049] 2. By performing linear interpolation on the detected projection image to obtain the interpolated restored projection image, the sampling density of the projection data is supplemented, reducing the impact of insufficient sampling density caused by the detected projection image. The interpolated restored projection image is optimized using the projection domain network model to significantly reduce the error caused by linear interpolation. The reconstructed image is further optimized in combination with the image domain network model, greatly improving the accuracy and quality of the reconstructed image.

[0050] 3. When performing linear interpolation on the detection projection image to increase the sampling density, the pixel value under the interpolation perspective is determined by using the interpolation perspective, the adjacent projection perspective, and the pixel value corresponding to the projection perspective, so that the interpolation is more natural and the error is smaller;

[0051] 4. In the projection domain network model, the interpolated restored projection image is extracted and predicted through the sequentially connected feature extraction module, upsampling module and splicing module to obtain an optimized projection image, which reduces the error of the interpolated restored projection image and improves the accuracy of the reconstructed image;

[0052] 5. Through the image domain network model with a dual encoder-single decoder network, multi-scale feature extraction and splicing fusion are performed on the optimized reconstructed image and the initial reconstructed image, so that the final target reconstructed image has higher quality.

[0053] In the present invention, the above-mentioned technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of the present invention will be described in the following description, and some advantages will become apparent from the description or be learned through practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like parts throughout the drawings.

[0055] Figure 1 A schematic structural diagram of a static CT acquisition system provided by an embodiment of the present invention;

[0056] Figure 2 This is a schematic diagram of a static data acquisition device according to an embodiment of the present invention;

[0057] Figure 3 A second structural diagram of a static data acquisition device provided by an embodiment of the present invention;

[0058] Figure 4 A third structural diagram of a static data acquisition device provided by an embodiment of the present invention;

[0059] Figure 5 One of the flowcharts of obtaining a target reconstructed image based on a detected projection image provided by an embodiment of the present invention;

[0060] Figure 6 A schematic diagram of a detection projection image provided by an embodiment of the present invention;

[0061] Figure 7 A schematic diagram of interpolating and restoring a projected image according to an embodiment of the present invention;

[0062] Figure 8 A schematic diagram of an optimized projection image provided by an embodiment of the present invention;

[0063] Figure 9 A schematic diagram of an optimized reconstructed image provided by an embodiment of the present invention;

[0064] Figure 10 A schematic diagram of the structure of a projection domain network model provided by an embodiment of the present invention;

[0065] Figure 11 A schematic structural diagram of a first deformable convolution module provided in an embodiment of the present invention;

[0066] Figure 12 A schematic diagram of the process of a multi-scale deformable convolution operation provided by an embodiment of the present invention;

[0067] Figure 13 A schematic diagram of the structure of a deformable convolution and deconvolution module provided in an embodiment of the present invention;

[0068] Figure 14 A schematic structural diagram of a second deformable convolution module provided in an embodiment of the present invention;

[0069] Figure 15 A schematic diagram of an initial reconstructed image provided by an embodiment of the present invention;

[0070] Figure 16 A schematic diagram of the structure of the image domain network model provided by an embodiment of the present invention;

[0071] Figure 17The second flowchart of the embodiment of the present invention is to reconstruct a target reconstructed image based on the detected projection image. DETAILED DESCRIPTION

[0072] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, and are not used to limit the scope of the present invention.

[0073] A specific embodiment of the present invention discloses a static CT acquisition system, such as Figure 1 As shown, the static CT acquisition system includes at least one static acquisition device arranged in sequence according to the object scanning order, and each static acquisition device has the same structure, including a base, a ray source array and a detector array;

[0074] The ray source array is arranged inside the base, and ray source windows are opened on the inner side of the base at every same preset area, and each ray source window corresponds to a ray source; wherein the distribution angle of the ray source array is greater than or equal to π+2γ and less than 2π, where γ represents the opening angle of the fan beam of the ray source;

[0075] The detector array includes a first detector sub-array, a second detector sub-array and a third detector sub-array; the first detector sub-array is arranged at each preset area inside the base, and the second detector sub-array and the third detector sub-array are arranged in other areas inside the base except the preset areas and the radiation source window.

[0076] Specifically, such as Figure 1 As shown, the present invention provides a static collection system including at least one static collection device arranged in sequence. When an object is in a forward direction, at least one static collection device is arranged in sequence to work. That is, after the first static collection device completes the scanning work, the next static collection device starts working, and so on, until all static collection devices in the static collection system complete the scanning work.

[0077] It is worth noting that the structures of the static acquisition devices included in the same static acquisition system are the same, such as Figure 2 As shown, they all include a base, a ray source array and a detector array. The ray source array includes multiple ray sources arranged in sequence, and the detector array includes several first detector sub-arrays, second detector sub-arrays and multiple third detector sub-arrays.

[0078] Specifically, such as Figure 2 、 Figure 3 and Figure 4As shown, the radiation source array is arranged inside the base, and a radiation source window is opened on the inner side of the base in each identical preset area. Each radiation source window corresponds to a radiation source on the radiation source array. A first detector sub-array is arranged in the preset area, and a third detector sub-array is arranged in the middle area between two adjacent first detector sub-arrays on the inner side of the base, excluding the radiation source window area, and a second detector sub-array is arranged in other areas inside the base.

[0079] Specifically, such as Figure 2 As shown, the central angle of the area occupied by the second detector sub-array is π-2γ, the central angle of the area occupied by the ray source window and the first detector sub-array is π+2γ, and as Figure 3 and Figure 4 As shown, the third detector sub-array is arranged in the remaining area between the two adjacent first detector sub-arrays excluding the area where the ray source window is located.

[0080] Preferably, the shape of the base is any one of the following:

[0081] Ring;

[0082] square;

[0083] Regular hexagon.

[0084] Specifically, when the shape of the base is annular, the manufacturing and layout of the base and the ray source array are more difficult, but less material is required; when the shape of the base is square or regular hexagonal, the manufacturing and layout of the base and the ray source array will be reduced, but at the same time more material is required. In actual use, a reasonable choice should be made according to the specific situation.

[0085] Preferably, the first detector sub-array, the second detector sub-array and the third detector sub-array are respectively connected to the base via fixing bolts.

[0086] Specifically, such as Figure 4 As shown, the first detector subarray is connected to the fixing bolt holes arranged on the base through fixing bolts. Similarly, the second detector subarray and the third detector subarray are both connected to the fixing bolt holes through fixing bolts.

[0087] Preferably, the static CT acquisition system further includes a ray source window controller;

[0088] The ray source window controller is used to control the working time of each ray source in all ray source arrays;

[0089] The detector array receives the ray source data and transmits the received ray source data as a detection projection image to the ray source window controller;

[0090] The ray source window controller is also used to reconstruct a target reconstructed image based on the detected projection image.

[0091] Specifically, the ray source window controller is used to control the working time of the ray source. In the same static CT acquisition system, the working time of each ray source is different. Figure 1 When the first static acquisition device is operating, each radiation source of the first static acquisition device is individually activated in chronological order to collect object detection data, while the radiation sources of the remaining static acquisition devices are all deactivated. When the second static acquisition device is operating, the radiation sources of the remaining static acquisition devices are all deactivated, and each radiation source of the second static acquisition device is individually activated in chronological order to collect object detection data. After all static acquisition devices in a static CT acquisition system have completed collecting object detection data, all data is used as the object's radiation source data.

[0092] It is worth noting that if Figure 2 、 Figure 3 and Figure 4 It can be seen from the static acquisition device shown that the collected ray source data is empty at the ray source window, so the area with empty data can be supplemented by the existing linear interpolation method, which will not be repeated here. The supplemented ray source data will be used as the ray source data of the object.

[0093] Specifically, the ray source data of the object is transmitted to the ray source window controller as the detected projection image of the object. The ray source window controller reconstructs the target reconstructed image according to the detected projection image, that is, reconstructs the reconstructed image of the object.

[0094] Preferably, if Figure 5 As shown, the target reconstructed image is obtained by reconstructing the detected projection image, including:

[0095] Step S41: Based on the linear interpolation method, the detected projection image is interpolated in the projection viewing angle direction to obtain an interpolated restored projection image; the interpolated restored projection image is input into the pre-trained projection domain network model to obtain an optimized projection image;

[0096] Step S42: reconstructing the optimized projection image and the detection projection image based on a preset image reconstruction algorithm to obtain an optimized reconstructed image and an initial reconstructed image;

[0097] Step S43: input the optimized reconstructed image and the initial reconstructed image into the pre-trained image domain network model to obtain the target reconstructed image.

[0098] Specifically, such as Figure 5As shown, all images acquired by the static CT acquisition system of at least one static acquisition device are arranged and spliced ​​in the order of projection viewing angles to obtain the following: Figure 6 The detection projection image shown, where Figure 6 The horizontal direction is the detector direction, and the vertical direction is the projection viewing angle direction. Figure 6 Each row in can be understood as the sampled image data at each projection perspective.

[0099] Specifically, such as Figure 5 As shown, in step S41, the detected projection image is linearly interpolated in the projection viewing angle direction to obtain an interpolated restored projection image, as shown in FIG. Figure 7 As shown, the resolution of the interpolated restored projection image in the projection viewing angle direction is improved, and the sampling data is improved.

[0100] Preferably, the interpolating the detected projection image in the projection viewing angle direction includes:

[0101] Determine one or more interpolation perspectives to be inserted between two adjacent projection perspectives, and determine the pixel points under each interpolation perspective;

[0102] Determine an interpolation pixel value of a pixel point under each interpolation viewing angle according to the interpolation viewing angle of the pixel point under each interpolation viewing angle, two adjacent projection viewing angles, and two adjacent projection pixel values ​​in the projection viewing angle direction;

[0103] The pixel points of all projection view angles and all interpolation view angles are arranged in the order of projection view angles to obtain the interpolated restored projection image.

[0104] Specifically, when performing interpolation, one or more interpolation perspectives can be set according to the size of the gap between two adjacent projection perspectives. It can be understood that the gap between any two adjacent projection perspectives may be different, so the number of interpolation perspectives required to be inserted between any two adjacent projection perspectives will be different.

[0105] Specifically, the number of pixels at each projection viewing angle is the same, and when linear interpolation is performed, the number of pixels at each interpolation viewing angle is made the same as the number of pixels at the projection viewing angle.

[0106] When calculating the interpolation pixel value of each pixel point under each interpolation viewing angle, two adjacent projection viewing angles and two corresponding projection pixel values ​​in the projection viewing angle direction are determined. Figure 6As shown, the first row represents the projection pixel values ​​under a 2° projection angle of view, each row includes 1000 pixels, and the second row represents the projection pixel values ​​under a 4° projection angle of view, each row includes 1000 pixels. Now, an interpolation angle needs to be inserted between the first and second rows. In this case, the interpolation angle of view can be set to 3°, so a row of pixels, i.e., 1000 rows, needs to be inserted under the 3° interpolation angle of view. When calculating the interpolation pixel values ​​of any column of pixels under the 3° interpolation angle of view, the interpolation pixel values ​​can be determined based on the projection pixel values ​​of the pixels of any column under the 2° projection angle of view and the projection pixel values ​​of the pixels of any column under the 4° projection angle of view.

[0107] Preferably, the interpolated pixel value of each pixel under the interpolation viewing angle is calculated by the following formula:

[0108]

[0109] Among them, β k and β k+1 Represents two adjacent projection perspectives, β k′ represents the interpolation perspective, P(β k′ , i) represents the interpolation viewing angle β k′ The interpolated pixel value of the i-th pixel under k , i) and P(β k+1 , i) represent the projection viewing angle β k and β k+1 The projected pixel value of the i-th pixel under .

[0110] Specifically, the above formula calculates the interpolated pixel value of any pixel point under the interpolated perspective by the angle difference between the interpolated perspective and two adjacent projection perspectives, so that the interpolated pixel point is closer to the real pixel value.

[0111] Specifically, such as Figure 7 As shown, after the interpolation pixel values ​​of all pixels under all interpolation viewing angles are calculated, the pixel points of all projection viewing angles and all interpolation viewing angles are arranged in the order of projection viewing angles to obtain the interpolated restored projection image.

[0112] Specifically, such as Figure 5 As shown, in step S41, the interpolated restored projection image is input into the pre-trained projection domain network model, and the optimized projection image is predicted by the projection domain network model, as shown in FIG. Figure 8As shown, the optimized projection image and the interpolated restored projection image have the same size. The difference is that the optimized projection image is an optimized interpolated restored projection image, and the interpolated pixels in the interpolated restored projection image can be corrected and optimized. The optimized projection image after correction and optimization is closer to the target object. In step S42, the optimized projection image is reconstructed by a preset image reconstruction algorithm, and the image quality of the obtained optimized reconstructed image is better, as shown in FIG. Figure 9 shown.

[0113] Preferably, if Figure 10 As shown, the projection domain network model includes a feature extraction module, an upsampling module and a splicing module connected in sequence;

[0114] The feature extraction module is used to extract multi-scale features from the input interpolated and restored projection image, and input the obtained feature maps of multiple scales into the upsampling module:

[0115] An upsampling module is used to perform layer-by-layer upsampling and splicing of multiple feature maps of different scales to obtain a tenth feature map, and output the obtained tenth feature map to the splicing module;

[0116] The splicing module is used to splice the input interpolated restored projection image and the tenth feature map, and use the spliced ​​feature map as the optimized projection image.

[0117] Specifically, such as Figure 10 As shown, the projection domain network model includes a feature extraction module, an upsampling module, and a splicing module. The feature extraction module is used to receive the interpolated restored projection image and perform multi-scale feature extraction on the interpolated restored projection image to obtain multiple first to fifth feature maps of different scales, which are respectively input into the upsampling module; the upsampling module performs layer-by-layer upsampling and splicing on the multiple first to fifth feature maps of different scales, and outputs the obtained tenth feature map P10 to the splicing module; the splicing module is used to splice the interpolated restored projection image and the tenth feature map P10, and use the spliced ​​feature map as the optimized projection image.

[0118] Preferably, if Figure 10 As shown, the feature extraction module includes a first deformable convolution module, a preset number of first dual multi-scale deformable convolution modules and a second dual multi-scale deformable convolution module connected in sequence;

[0119] A first deformable convolution module is used to perform a deformable convolution operation and a multi-scale deformable convolution operation on the input interpolated restored projection image to obtain a first feature map;

[0120] A first dual multi-scale deformable convolution module performs a dual multi-scale deformable convolution operation on the first feature map to obtain a second feature map;

[0121] Multiple intermediate first dual multi-scale deformable convolution modules respectively perform dual multi-scale deformable convolution operations on the feature maps output by the previous first dual multi-scale deformable convolution module to obtain feature maps of different scales as multiple third feature maps;

[0122] The last first dual multi-scale deformable convolution module performs a dual multi-scale deformable convolution operation on the feature map output by the previous first dual multi-scale deformable convolution module to obtain a fourth feature map;

[0123] The second dual multi-scale deformable convolution module is used to perform a dual multi-scale deformable convolution operation on the fourth feature map to obtain a fifth feature map.

[0124] Specifically, such as Figure 10 As shown, after the interpolated restored projected image is input into the feature extraction module, the obtained feature map passes through the first deformable convolution module, a preset number of first dual multi-scale deformable convolution modules and the second dual multi-scale deformable convolution module in sequence to obtain the first feature map P1, the second feature map P2, multiple third feature maps P3, the fourth feature map P4 and the fifth feature map P5 respectively.

[0125] It is worth noting that the preset number is set in advance, and the preset number is greater than or equal to 3, that is, the feature extraction module includes at least 3 first dual multi-scale deformable convolution modules.

[0126] Specifically, such as Figure 10 As shown, the preset number of first dual multi-scale deformable convolution modules includes a first first dual multi-scale deformable convolution module, multiple intermediate first dual multi-scale deformable convolution modules and a last first dual multi-scale deformable convolution module connected in sequence, wherein the feature map output by the first first dual multi-scale deformable convolution module is used as the second feature map P2, the feature map output by the intermediate first dual multi-scale deformable convolution module is used as the third feature map P3, and the feature map output by the last first dual multi-scale deformable convolution module is used as the fourth feature map P4.

[0127] It is worth noting that when the feature extraction module includes three or more first dual multi-scale deformable convolution modules, the feature extraction module outputs multiple third feature maps P3. It can be understood that the sizes of the multiple third feature maps P3 are different.

[0128] Preferably, if Figure 11 As shown, the first deformable convolution module includes a deformable convolution module and a multi-scale deformable convolution module connected in sequence;

[0129] The first dual multi-scale deformable convolution module and the second dual multi-scale deformable convolution module have the same structure, and both include two multi-scale deformable convolution modules with the same structure;

[0130] The multi-scale deformable convolution module performs multi-scale deformable convolution operations, batch normalization operations, and ReLU activation function operations on the input feature map.

[0131] Specifically, the first dual multi-scale deformable convolution module and the second dual multi-scale deformable convolution module have the same structure, and both include two multi-scale deformable convolution modules with the same structure connected in sequence.

[0132] Specifically, the deformable convolution module performs a deformable convolution operation on the input feature map, and the multi-scale deformable convolution module performs a multi-scale deformable convolution operation, a batch normalization operation, and a ReLU activation function operation on the input feature map.

[0133] Specifically, such as Figure 10 As shown, the first deformable convolution module performs a deformable convolution operation and a multi-scale deformable convolution operation on the input interpolated restored projection image to obtain a first feature map P1.

[0134] Preferably, the multi-scale deformable convolution operation includes the following steps:

[0135] Perform channel separation on the input feature map to obtain feature maps of multiple different channels;

[0136] The feature maps of multiple different channels are subjected to convolution operations of different scales to obtain feature maps of multiple different scales;

[0137] Channel splicing is performed on multiple feature maps of different scales, and the obtained feature map is spliced ​​with the input feature map as the output feature map.

[0138] Specifically, such as Figure 12 As shown in the figure, the input feature map is firstly channel-separated in the multi-scale deformable convolution operation, and the input feature map is divided into feature maps of multiple different channels. Secondly, the feature maps of multiple different channels are subjected to convolution operations of different scales, such as 3×3 deformable convolution operation, 1×11 deformable convolution operation, 11×1 deformable convolution operation and identity mapping operation to obtain feature maps of multiple scales. Finally, the feature maps of multiple scales are spliced ​​through channels and output as the output feature map.

[0139] Preferably, if Figure 10 As shown, the upsampling module includes a two-dimensional deconvolution module, a preset number of deformable convolution and deconvolution modules, and a second deformable convolution module;

[0140] The two-dimensional deconvolution module performs a two-dimensional deconvolution operation on the input fifth feature map to obtain a sixth feature map;

[0141] The sixth feature map is concatenated with the fourth feature map of the same scale and input into the first deformable convolution and deconvolution module. Each deformable convolution and deconvolution module is used to perform a deformable convolution operation and a two-dimensional deconvolution operation on the input feature map, and output feature maps of different scales.

[0142] The input of each deformable convolution and deconvolution module is the feature map obtained by concatenating the feature map output by the previous deformable convolution and deconvolution module and the third feature map of the same scale output by the feature extraction module;

[0143] The input of the last deformable convolution and deconvolution module is the concatenation of the feature map output by the previous deformable convolution and deconvolution module and the second feature map of the same scale;

[0144] The feature map output by the last deformable convolution and deconvolution module is concatenated with the first feature map of the same scale and input into the second deformable convolution module to obtain the tenth feature map.

[0145] Specifically, the upsampling module includes a two-dimensional deconvolution module, a preset number of deformable convolution and deconvolution modules, and a second deformable convolution module. The preset number is set in advance, and the number of the first dual multi-scale deformable convolution module and the deformable convolution and deconvolution module needs to be the same.

[0146] like Figure 10 As shown, the fifth feature map P5 undergoes a two-dimensional deconvolution operation in the two-dimensional deconvolution module to obtain a sixth feature map P6. The sixth feature map P6 and the fourth feature map P4 are spliced ​​and input into the first deformable convolution and deconvolution module. The first deformable convolution and deconvolution module performs a deformable convolution operation and a two-dimensional deconvolution operation on the input feature map to obtain the seventh feature map P7; the seventh feature map P7 and the third feature map P3 of the same scale are spliced ​​and input into the middle deformable convolution and deconvolution module. The feature map output by the middle deformable convolution and deconvolution module is used as the eighth feature map P8. The feature map output by the last deformable convolution and deconvolution module is used as the ninth feature map P9. The ninth feature map P9 and the first feature map P1 are spliced ​​and input into the second deformable convolution module to obtain the tenth feature map P10.

[0147] Preferably, if Figure 13 As shown, the deformable convolution and deconvolution module includes a deformable convolution module and a two-dimensional deconvolution module connected in sequence;

[0148] The deformable convolution module performs deformable convolution, batch normalization, and ReLU activation function operations on the input feature map in sequence;

[0149] The two-dimensional deconvolution module performs a two-dimensional deconvolution operation on the input feature map.

[0150] Specifically, such as Figure 13 As shown in the figure, the feature map input to the deformable convolution and deconvolution module passes through the deformable convolution module and the two-dimensional deconvolution module in sequence, and performs deformable convolution operation, batch normalization operation and ReLU activation function operation, as well as two-dimensional deconvolution operation respectively to obtain the output feature map.

[0151] Preferably, if Figure 14 As shown, the second deformable convolution module includes a deformable convolution module and a common convolution module connected in sequence;

[0152] The ordinary convolution module performs convolution operation, batch normalization operation and ReLU activation function operation on the input feature map to obtain the tenth feature map.

[0153] Specifically, such as Figure 14 As shown, the feature map input into the second deformable convolution module passes through the deformable convolution module and the ordinary convolution module in sequence, and performs deformable convolution operation, batch normalization operation and ReLU activation function operation, as well as convolution operation, batch normalization operation and ReLU activation function operation respectively to obtain the output feature map as the tenth feature map P10.

[0154] It can be understood that the interpolated restored projection image input into the feature extraction module undergoes feature extraction in sequence, and its receptive field can dynamically adapt to changes in local projection features, thereby improving the modeling ability of the projection domain network model in irregular projections and obtaining higher quality images; at the same time, the feature maps of each scale obtained in the feature extraction module are spliced ​​with the feature maps of the same scale generated in the upsampling module, and then operated again, which can not only retain the original information but also enhance the feature expression ability, so that the obtained optimized projection image can significantly reduce the pixel value error caused by linear interpolation compared to the interpolated restored projection image.

[0155] Specifically, such as Figure 5 As shown, in step S42, the optimized projection image is reconstructed based on a preset image reconstruction algorithm to obtain an optimized reconstructed image, such as Figure 9 As shown, the detection projection image is reconstructed based on the preset image reconstruction algorithm to obtain the initial reconstructed image, as shown in Figure 15 shown.

[0156] Exemplarily, the preset image reconstruction algorithm is any of the following reconstruction algorithms:

[0157] Filtered back projection algorithm;

[0158] Convolution back-projection algorithm;

[0159] Statistical iterative reconstruction algorithm;

[0160] Algebraic reconstruction algorithm;

[0161] Reconstruction algorithm based on compressed sensing.

[0162] Specifically, such as Figure 5 As shown, in step S43, the optimized reconstructed image and the initial reconstructed image are input into the image domain network model, and the obtained result is used as the target reconstructed image.

[0163] Preferably, if Figure 16 As shown, the image domain network model is a dual encoder-single decoder network, which includes a first feature extraction module and a second feature extraction module with the same structure, as well as a feature fusion module;

[0164] The first feature extraction module and the second feature extraction module are used to perform multi-scale feature extraction on the optimized reconstructed image and the initial reconstructed image, respectively, to obtain a multi-scale optimized feature map and a multi-scale initial feature map;

[0165] The feature fusion module is used to splice and fuse the optimized feature map and the initial feature map of the same scale to obtain the target reconstructed image.

[0166] Preferably, the dual encoder-single decoder network is composed of any of the following networks:

[0167] Convolutional neural networks;

[0168] Self-attention networks;

[0169] Feedforward neural network.

[0170] Specifically, such as Figure 16 As shown, in the image domain network model, the first feature extraction module and the second feature extraction module are used as dual encoders, and the feature fusion module is used as a single decoder. The first feature extraction module performs multi-scale feature extraction on the optimized reconstructed image to obtain a multi-scale optimized feature map. The second feature extraction module performs multi-scale feature extraction on the initial reconstructed image to obtain a multi-scale initial feature map. The feature fusion module splices and fuses the multi-scale optimized feature map and the multi-scale initial feature map to obtain the target reconstructed image.

[0171] Preferably, if Figure 16 As shown, the first feature extraction module and the second feature extraction module have the same structure, both including a plurality of dual convolution modules connected in sequence; the dual convolution module includes two convolution modules connected in sequence, each convolution module is used to perform convolution operation, normalization operation and ReLU activation function operation on the input feature map;

[0172] The first dual convolution module is used to receive the optimized reconstructed image or the initial reconstructed image, and transmit the obtained feature map to the next dual convolution module and output it to the feature fusion module;

[0173] The intermediate dual convolution module is used to obtain the feature map of the current dual convolution module based on the feature map output by the previous dual convolution module, and transmit the feature map of the current dual convolution module to the next dual convolution module and output it to the feature fusion module;

[0174] The last double convolution module is used to obtain the feature map of the last double convolution module according to the feature map output by the previous double convolution module, and output the feature map of the last double convolution module to the feature fusion module.

[0175] Specifically, the dual convolution module includes two convolution modules connected in sequence, and each convolution module performs convolution operation, normalization operation and ReLU activation function operation on the input feature map.

[0176] Specifically, the optimized reconstructed image is input into the first feature extraction module, and after passing through multiple double convolution modules in sequence, each double convolution module outputs an optimized feature map of one scale; similarly, the initial reconstructed image is input into the second feature extraction module, and each double convolution module outputs an initial feature map of one scale; and the multi-scale optimized feature map and the multi-scale initial feature map are output to the feature fusion module.

[0177] Preferably, the feature fusion module includes a spliced ​​two-dimensional deconvolution module, a plurality of spliced ​​convolution and two-dimensional deconvolution modules and a spliced ​​double convolution splicing module connected in sequence;

[0178] A splicing two-dimensional deconvolution module is used to receive the feature map of the last double convolution module in the first feature extraction module and the second feature extraction module, and perform splicing and two-dimensional deconvolution operations, and output the obtained feature map to the first splicing convolution and two-dimensional deconvolution module; each splicing convolution and two-dimensional deconvolution module is used to perform splicing, convolution, normalization, ReLU activation function and two-dimensional deconvolution on the input feature map;

[0179] The first splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the splicing two-dimensional deconvolution module, as well as the feature maps output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the splicing two-dimensional deconvolution module, and transmit the obtained feature map to the next splicing convolution and two-dimensional deconvolution module;

[0180] The intermediate splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the previous splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the previous splicing convolution and two-dimensional deconvolution module, and transmit the processed feature map to the next splicing convolution and two-dimensional deconvolution module;

[0181] The last splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the previous splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the previous splicing convolution and two-dimensional deconvolution module, and transmit the obtained feature map to the splicing double convolution splicing module;

[0182] The splicing double convolution splicing module is used to receive the feature map output by the last splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the last splicing convolution and two-dimensional deconvolution module, and perform splicing operation, convolution operation, normalization operation, ReLU activation function operation, convolution operation, normalization operation and ReLU activation function operation on the input feature map in sequence, and perform splicing operation on the obtained feature map and the optimized reconstructed image as the target reconstructed image.

[0183] Specifically, such as Figure 16 As shown in the figure, in the splicing two-dimensional deconvolution module, the input feature map is spliced, and the feature map obtained by the splicing operation is then subjected to a two-dimensional deconvolution operation. The final feature map is output to the first splicing convolution and two-dimensional deconvolution module.

[0184] Specifically, the splicing two-dimensional deconvolution module includes a splicing module and a two-dimensional deconvolution module connected in sequence. The feature map of the last double convolution module in the first feature extraction module and the second feature extraction module input into the splicing two-dimensional deconvolution module is first spliced ​​through the splicing module, and the obtained feature map is then input into the two-dimensional deconvolution module for two-dimensional deconvolution operation. Finally, the obtained feature map is input into the first splicing convolution and two-dimensional deconvolution modules.

[0185] Specifically, such as Figure 16 As shown in the figure, in the splicing convolution and two-dimensional deconvolution module, the input feature map is sequentially subjected to splicing operation, convolution operation, normalization operation, ReLU activation function operation and two-dimensional deconvolution operation, and the final feature map is output.

[0186] Specifically, each splicing convolution and two-dimensional deconvolution module includes a splicing module, a convolution module and a two-dimensional deconvolution module connected in sequence. The feature map input to the splicing convolution and two-dimensional deconvolution module first passes through the splicing module for splicing operation, and then is input into the convolution module for convolution operation, normalization operation and ReLU activation function operation, and then input into the two-dimensional deconvolution module for two-dimensional deconvolution operation, and finally output.

[0187] It can be understood that the final feature map obtained in the intermediate splicing convolution and two-dimensional deconvolution module is output to the next splicing convolution and two-dimensional deconvolution module, and the final feature map obtained in the last splicing convolution and two-dimensional deconvolution module is output to the splicing double convolution splicing module.

[0188] It is worth noting that in the splicing double convolution splicing module, on the one hand, the feature map output by the last splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the last splicing convolution and two-dimensional deconvolution module are first spliced ​​together, and then the feature map after the splicing operation is subjected to convolution operation, normalization operation, ReLU activation function operation, convolution operation, normalization operation and ReLU activation function operation, and finally the obtained feature map is spliced ​​with the input optimized reconstructed image as the target reconstructed image.

[0189] Specifically, the splicing double convolution splicing module includes a first splicing module, a first convolution module, a second convolution module and a second splicing module connected in sequence; in the first splicing module, the feature map output by the last splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the last splicing convolution and two-dimensional deconvolution module are received, and a splicing operation is performed, and the feature map obtained after the splicing operation is output to the first convolution module; in the first convolution module, a convolution operation, a normalization operation and a ReLU activation function operation are performed on the input feature map, and the obtained feature map is output to the second convolution module; in the second convolution module, a convolution operation, a normalization operation and a ReLU activation function operation are performed on the input feature map, and the obtained feature map is output to the second splicing module; in the second splicing module, the optimized reconstructed image and the feature map output by the second convolution module are received, and a splicing operation is performed, and the feature map obtained after the splicing operation is used as the target reconstructed image.

[0190] It is understandable that in order to effectively train the projection domain network model and the image domain network model, it is necessary to construct the following paired datasets, which require two paired datasets: full sampling (high quality) and sparse sampling (low quality).

[0191] For example, the training data set construction method and network training process are as follows:

[0192] 1. Data Collection

[0193] 1. High-quality fully sampled data (Ground Truth)

[0194] Projection data:

[0195] Total number of projection angles: N ≥ 720 (N must be a multiple of N', where N' is the number of projection angles);

[0196] Acquisition system: spiral CT;

[0197] The geometric parameters (source-detector distance, pixel size, etc.) need to remain consistent with those of the static CT system.

[0198] Reconstruct the image:

[0199] Image size: 512 × 512 pixels;

[0200] Reconstruction algorithm: filtered back projection algorithm is used.

[0201] 2. Sparsely sampled low-quality data (input)

[0202] Projection data:

[0203] The fully sampled data is downsampled to the corresponding angle projection of the static CT system, with the sampling angle being N'. At the same time, the projection data of the theoretical missing area needs to be set to zero to simulate the actual data missing situation;

[0204] Or directly scan the same object using a static CT system.

[0205] Reconstruct the image:

[0206] Keep the same reconstruction parameters (algorithm, image size, pixel size, etc.) as the fully sampled data.

[0207] 2. Dataset Construction Specifications

[0208] Data scale:

[0209] CT scan data of no less than 100 different objects;

[0210] Each object must contain 300 scan layers;

[0211] A total of 30,000 images of 512 × 512 pixels were generated.

[0212] Data preprocessing:

[0213] Normalizing the projection data and reconstructed image data;

[0214] Set the value range to compress the pixel value to the interval [0,1].

[0215] Data partitioning:

[0216] Divide in a ratio of 7:1:2;

[0217] Training set (70%): used for model training and parameter updating;

[0218] Validation set (10%): used for real-time testing and overfitting monitoring during training;

[0219] Test set (20%): used for final model performance evaluation.

[0220] 3. Network Training

[0221] Phase 1:

[0222] Input: sparsely sampled low-quality projection data;

[0223] Processing: Perform preliminary interpolation repair;

[0224] Output: Serves as input to the projection domain network model.

[0225] Phase 2:

[0226] Training target: Projection domain network model;

[0227] Input: projection after interpolation repair;

[0228] Supervision signal: high-quality fully sampled projection;

[0229] Loss function: Among them, P pred is the optimized projection of the network output, P gt are high-quality fully sampled projections used as supervision.

[0230] Phase 3:

[0231] Training target: image domain network model;

[0232] Input: optimized projected reconstructed image + original sparse sampling reconstructed image;

[0233] Supervisory signal: high-quality CT images reconstructed from fully sampled projections;

[0234] Loss function: Among them I pred is the reconstructed CT output by the network, I gt High-quality CT images reconstructed from fully sampled projections used as supervision.

[0235] 4. Training strategy:

[0236] Adopt phased training: first complete the projection domain network training, then conduct image domain network training;

[0237] Each network was trained for 200 rounds;

[0238] Initial learning rate: 0.0005;

[0239] Learning rate adjustment: decays to 70% of the previous value every 20 epochs.

[0240] It can be understood that after the projection domain network model and the image domain network model are trained and applied, the detection projection image is reconstructed through the dual network model, and the quality of the target reconstructed image is still very high and the distortion is low.

[0241] The target reconstructed image is obtained by using the projection domain network model and the image domain network model, for example, Figure 17 As shown, the detection projection image is restored by interpolation to obtain an interpolated restored projection image, the interpolated restored projection image is input into the projection domain network model to obtain an optimized projection image, the optimized projection image and the detection projection image are reconstructed respectively to obtain an optimized reconstructed image and an initial reconstructed image, the optimized reconstructed image and the initial reconstructed image are simultaneously input into the image domain network model to obtain a target reconstructed image, thereby obtaining a target reconstructed image based on the detection projection image, which greatly reduces artifacts and improves the quality of the reconstructed image.

[0242] Compared with the prior art, an image reconstruction method based on a dual network model provided by an embodiment of the present invention arranges a ray source array inside a base, opens a ray source window inside the base at every same preset area, arranges a first detector subarray at each preset area inside the base, and arranges a second detector subarray and a third detector subarray in other areas inside the base except the preset area and the ray source window. When in use, the working time of each ray source is controlled, thereby reducing the manufacturing difficulty and cost of the CT acquisition equipment, so that the entire static acquisition device does not need to make additional movements during operation, thereby improving the acquisition efficiency of the object; by performing linear interpolation on the detected projection image to obtain an interpolated restored projection image to additionally supplement the sampling density of the projection data, the influence of insufficient sampling density caused by the detected projection image is reduced, and the projection domain network model is used to optimize the interpolated restored projection image. The error caused by linear interpolation is significantly reduced, and the reconstructed image is further optimized in combination with the image domain network model, which greatly improves the accuracy and quality of the reconstructed image. At the same time, when linear interpolation is performed on the detected projection image to increase the sampling density, the pixel value under the interpolation perspective is determined by using the interpolation perspective, the adjacent projection perspective and the pixel value corresponding to the projection perspective, which makes the interpolation more natural and the error smaller. Moreover, in the projection domain network model, the interpolated restored projection image is feature extracted and predicted through the sequentially connected feature extraction module, upsampling module and splicing module to obtain the optimized projection image, which reduces the error of the interpolated restored projection image and improves the accuracy of the reconstructed image. Finally, the optimized reconstructed image and the initial reconstructed image are subjected to multi-scale feature extraction and splicing fusion through the image domain network model with a dual encoder-single decoder network, so that the final target reconstructed image has higher quality.

[0243] Those skilled in the art will appreciate that all or part of the process steps of the above-described embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, such as a magnetic disk, an optical disk, a read-only memory, or a random access memory.

[0244] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A static CT acquisition system, characterized in that: The static CT acquisition system includes at least one static acquisition device arranged in sequence according to the object scanning order, each static acquisition device has the same structure and includes a base, a ray source array and a detector array; The ray source array is arranged inside the base, and ray source windows are opened on the inner side of the base at every same preset area, and each ray source window corresponds to a ray source; wherein the distribution angle of the ray source array is greater than or equal to π+2γ and less than 2π, where γ represents the opening angle of the fan beam of the ray source; The detector array includes a first detector sub-array, a second detector sub-array and a third detector sub-array; the first detector sub-array is arranged at each preset area inside the base, and the second detector sub-array and the third detector sub-array are arranged in other areas inside the base except the preset areas and the radiation source window.

2. The static CT acquisition system according to claim 1, characterized in that: The shape of the base can be any of the following: Ring; square; Regular hexagon.

3. The static CT acquisition system according to claim 1, characterized in that: The first detector sub-array, the second detector sub-array and the third detector sub-array are respectively connected to the base through fixing bolts.

4. The static CT acquisition system according to any one of claims 1 to 3, characterized in that: The static CT acquisition system also includes a ray source window controller; The ray source window controller is used to control the working time of each ray source in all ray source arrays; The detector array receives the ray source data and transmits the received ray source data as a detection projection image to the ray source window controller; The ray source window controller is also used to reconstruct a target reconstructed image based on the detected projection image.

5. The static CT acquisition system according to claim 4, characterized in that: The step of reconstructing a target reconstructed image based on the detected projection image includes: Based on the linear interpolation method, the detected projection image is interpolated in the projection viewing angle direction to obtain the interpolated restored projection image; the interpolated restored projection image is input into the pre-trained projection domain network model to obtain the optimized projection image; The optimized projection image and the detection projection image are reconstructed based on a preset image reconstruction algorithm to obtain an optimized reconstructed image and an initial reconstructed image; The optimized reconstructed image and the initial reconstructed image are input into the pre-trained image domain network model to obtain the target reconstructed image.

6. The static CT acquisition system according to claim 5, characterized in that: The image domain network model is a dual encoder-single decoder network, which includes a first feature extraction module and a second feature extraction module with the same structure, and a feature fusion module; The first feature extraction module and the second feature extraction module are used to perform multi-scale feature extraction on the optimized reconstructed image and the initial reconstructed image, respectively, to obtain a multi-scale optimized feature map and a multi-scale initial feature map; The feature fusion module is used to splice and fuse the optimized feature map and the initial feature map of the same scale to obtain the target reconstructed image; The first feature extraction module and the second feature extraction module have the same structure, and both include a plurality of dual convolution modules connected in sequence; the dual convolution module includes two convolution modules connected in sequence, each convolution module is used to perform a convolution operation, a normalization operation and a ReLU activation function operation on the input feature map; The first dual convolution module is used to receive the optimized reconstructed image or the initial reconstructed image, and transmit the obtained feature map to the next dual convolution module and output it to the feature fusion module; the middle dual convolution module is used to obtain the feature map of the current dual convolution module based on the feature map output by the previous dual convolution module, and transmit the feature map of the current dual convolution module to the next dual convolution module and output it to the feature fusion module; The last double convolution module is used to obtain the feature map of the last double convolution module according to the feature map output by the previous double convolution module, and output the feature map of the last double convolution module to the feature fusion module.

7. The static CT acquisition system according to claim 6, characterized in that: The feature fusion module includes a spliced ​​two-dimensional deconvolution module, a plurality of spliced ​​convolution and two-dimensional deconvolution modules and a spliced ​​double convolution splicing module connected in sequence; A splicing two-dimensional deconvolution module is used to receive the feature map of the last double convolution module in the first feature extraction module and the second feature extraction module, and perform splicing and two-dimensional deconvolution operations, and output the obtained feature map to the first splicing convolution and two-dimensional deconvolution module; each splicing convolution and two-dimensional deconvolution module is used to perform splicing, convolution, normalization, ReLU activation function and two-dimensional deconvolution on the input feature map; The first splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the splicing two-dimensional deconvolution module, as well as the feature maps output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the splicing two-dimensional deconvolution module, and transmit the obtained feature map to the next splicing convolution and two-dimensional deconvolution module; The intermediate splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the previous splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the previous splicing convolution and two-dimensional deconvolution module, and transmit the obtained feature map to the next splicing convolution and two-dimensional deconvolution module; The last splicing convolution and two-dimensional deconvolution module is used to receive the feature map output by the previous splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the previous splicing convolution and two-dimensional deconvolution module, and transmit the obtained feature map to the splicing double convolution splicing module; The splicing double convolution splicing module is used to receive the feature map output by the last splicing convolution and two-dimensional deconvolution module, as well as the feature map output by the first feature extraction module and the second feature extraction module with the same scale as the feature map output by the last splicing convolution and two-dimensional deconvolution module, and perform splicing operation, convolution operation, normalization operation, ReLU activation function operation, convolution operation, normalization operation and ReLU activation function operation on the input feature map in sequence, and perform splicing operation on the obtained feature map and the optimized reconstructed image as the target reconstructed image.

8. The static CT acquisition system according to claim 5, characterized in that: The projection domain network model includes a feature extraction module, an upsampling module and a splicing module connected in sequence; The feature extraction module is used to extract multi-scale features from the input interpolated and restored projection image, and input the obtained feature maps of multiple scales into the upsampling module: An upsampling module is used to perform layer-by-layer upsampling and splicing of multiple feature maps of different scales to obtain a tenth feature map, and output the obtained tenth feature map to the splicing module; The splicing module is used to splice the input interpolated restored projection image and the tenth feature map, and use the spliced ​​feature map as the optimized projection image.

9. The static CT acquisition system according to claim 8, characterized in that: The feature extraction module includes a first deformable convolution module, a preset number of first dual multi-scale deformable convolution modules and a second dual multi-scale deformable convolution module connected in sequence; A first deformable convolution module is used to perform a deformable convolution operation and a multi-scale deformable convolution operation on the input interpolated restored projection image to obtain a first feature map; The first first dual multi-scale deformable convolution module performs a dual multi-scale deformable convolution operation on the first feature map to obtain a second feature map; multiple intermediate first dual multi-scale deformable convolution modules respectively perform a dual multi-scale deformable convolution operation on the feature map output by the previous first dual multi-scale deformable convolution module to obtain feature maps of different scales as multiple third feature maps; the last first dual multi-scale deformable convolution module performs a dual multi-scale deformable convolution operation on the feature map output by the previous first dual multi-scale deformable convolution module to obtain a fourth feature map; The second dual multi-scale deformable convolution module is used to perform a dual multi-scale deformable convolution operation on the fourth feature map to obtain a fifth feature map.

10. The static CT acquisition system according to claim 8, characterized in that: The upsampling module includes a two-dimensional deconvolution module, a preset number of deformable convolution and deconvolution modules, and a second deformable convolution module; The two-dimensional deconvolution module performs a two-dimensional deconvolution operation on the input fifth feature map to obtain a sixth feature map; The sixth feature map is concatenated with the fourth feature map of the same scale and input into the first deformable convolution and deconvolution module. Each deformable convolution and deconvolution module is used to perform a deformable convolution operation and a two-dimensional deconvolution operation on the input feature map, and output feature maps of different scales. The input of each deformable convolution and deconvolution module is the feature map obtained by concatenating the feature map output by the previous deformable convolution and deconvolution module and the third feature map of the same scale output by the feature extraction module; The input of the last deformable convolution and deconvolution module is the concatenation of the feature map output by the previous deformable convolution and deconvolution module and the second feature map of the same scale; The feature map output by the last deformable convolution and deconvolution module is concatenated with the first feature map of the same scale and input into the second deformable convolution module to obtain the tenth feature map.

Images