Dual-tracer ECT imaging method, device, equipment, and storage medium

Abstract

The present invention discloses a dual-tracer ECT imaging method, apparatus, device, and storage medium. The method comprises: acquiring first projection data of a target acquired by a high-energy ray detector, second projection data of the target acquired by a low-energy ray detector, and MR image data of the target; inputting the first projection data and the MR image data into a pre-trained PET imaging model, and performing image reconstruction using the MR image data as prior data to obtain a PET image; and constructing a SPECT image using the second projection data using the PET image in a nuclear feature-guided manner. The present invention can simultaneously create PET and SPECT images, and uses the PET image to create the SPECT image, resulting in higher-quality SPECT images.

Description

Technical Field

[0001] The present application relates to the technical field of medical image reconstruction, and in particular to a dual-tracer ECT imaging method, apparatus, device, and storage medium. Background Art

[0002] Medical imaging technologies are divided into two categories: structural imaging and functional imaging. Common structural imaging technologies include CT (Computed Tomography), MRI (Magnetic Resonance Imaging), and ultrasound imaging, while functional imaging technologies include optical and nuclear medicine functional imaging lamps. Nuclear medicine functional imaging is divided into two imaging technologies, PET and SPECT, based on the different radioactive tracers in the body. The former uses positron-emitting radioactive tracers labeled with positron nuclides, while the latter uses single-photon emitting tracers labeled with single-photon nuclides. Traditional PET and SPECT dual-mode imaging systems mostly use two different sets of detectors to obtain PET high-energy gamma ray and SPECT low-energy gamma ray data, respectively. They have a high radiation dose to the human body, are bulky, costly, and have poor mobility and adjustability. In addition, existing SPECT imaging quality is often poor, with low resolution and sensitivity. Summary of the Invention

[0003] The present application provides a dual-tracer ECT imaging method, apparatus, device and storage medium to solve the problem of single medical imaging method and poor imaging quality.

[0004] To solve the above technical problems, a technical solution adopted in the present application is: to provide a dual-tracer ECT imaging method, comprising: obtaining first projection data of a target acquired by a high-energy ray detector and second projection data of the target acquired by a low-energy ray detector, as well as MR image data of the target; inputting the first projection data and the MR image data into a pre-trained PET imaging model, and performing image reconstruction using the MR image data as prior data to obtain a PET image, wherein the PET imaging model is trained based on pre-prepared sample PET projection data and sample MR image data aligned with the sample PET projection data; and using the PET image to guide the second projection data to construct a SPECT image based on nuclear feature guidance.

[0005] As a further improvement of the present application, the PET imaging model is expressed as:

[0006]

[0007] x=f(θ|z);

[0008] Wherein, θ is the neural network parameter of the PET imaging model, L(y|θ) is the log-likelihood function representation of the first projection data, M is the total number of annihilation photon coincidence lines in the PET detector, and y i is the i-th first projection data, x is the PET image to be reconstructed, f represents the neural network of the PET imaging model, and z is the MR image data;

[0009] The maximum likelihood estimation method is used to reconstruct the PET image, which is expressed as:

[0010]

[0011]

[0012] in, is the maximum likelihood estimate of the neural network parameters, is the maximum likelihood estimate of the PET image to be reconstructed.

[0013] As a further improvement of the present application, the method based on nuclear feature guidance uses the PET image to guide the second projection data to construct a SPECT image, including:

[0014] Extracting all features of the PET image and constructing a kernel matrix based on all the features;

[0015] Iteratively calculating an optimal coefficient image estimation value using a maximum likelihood estimation method based on the kernel matrix and the second projection data;

[0016] The SPECT image is reconstructed using the optimal numerical image estimate and the kernel matrix.

[0017] As a further improvement of the present application, the expression of the optimal coefficient image estimation value is:

[0018]

[0019] in, is the optimal coefficient image estimation value, b is the second projection data, is the kernel matrix, α is the coefficient image estimation value;

[0020] The expression of the optimal coefficient image estimation value is iterated using the maximum EM algorithm combined with Poisson random distribution and is expressed as:

[0021]

[0022] Where R is the probability matrix of the relationship between the nuclide distribution and the detection projection of the SPECT system, c represents the random scattering event vector, T represents the matrix transpose, 1 Nis a vector of length N whose elements are all 1, and n is the number of iterations.

[0023] As a further improvement of the present application, extracting all features of the PET image and constructing a kernel matrix based on all features includes:

[0024] Extracting all feature data of the PET image using a radial Gaussian kernel function;

[0025] Based on the nearest neighbor search method, a sparse matrix is ​​established according to all the feature data;

[0026] Regularizing the sparse matrix to obtain the kernel matrix.

[0027] As a further improvement of the present application, the extraction of the feature data is expressed as follows:

[0028]

[0029] Wherein, k is the radial Gaussian kernel function, f j is the jth pixel in the PET image, f l f j One of the neighboring pixels of , σ is a preset parameter used to adjust the difference between neighboring pixels;

[0030] The establishment of the sparse matrix is ​​expressed as:

[0031]

[0032] Wherein, K is the sparse matrix;

[0033] The sparse matrix regularization is expressed as:

[0034]

[0035] in, is a regularized sparse matrix, diag -1 is the inverse matrix of the diagonal matrix, 1 N is a vector of length N whose elements are all 1, the same size as the rows and columns of the sparse matrix.

[0036] As a further improvement of the present application, the PET imaging model is implemented based on the 3DU-net network structure; training the PET imaging model includes: performing a preset number of MLEM reconstructions using sample PET projection data to obtain a sample reconstructed image; labeling the sample PET projection data using the sample reconstructed image; and iteratively training the PET imaging model using the labeled sample PET projection data and the sample MR image data aligned with the sample PET projection data.

[0037] To solve the above technical problems, another technical solution adopted in the present application is: to provide a dual-tracer ECT imaging device, including: an acquisition module, used to acquire first projection data of the target acquired by a high-energy ray detector and second projection data of the target acquired by a low-energy ray detector, as well as MR image data of the target; a PET imaging module, used to input the first projection data and the MR image data into a pre-trained PET imaging model, and perform image reconstruction using the MR image data as prior data to obtain a PET image, wherein the PET imaging model is trained based on pre-prepared sample PET projection data and sample MR image data aligned with the sample PET projection data; and a SPECT imaging module, used to use the PET image to guide the second projection data to construct a SPECT image based on a nuclear feature-guided manner.

[0038] To solve the above technical problems, another technical solution adopted in the present application is: providing a computer device, wherein the computer device includes a processor and a memory coupled to the processor, wherein program instructions are stored in the memory, and when the program instructions are executed by the processor, the processor executes the steps of the dual-tracer ECT imaging method as described in any one of the above items.

[0039] In order to solve the above technical problems, another technical solution adopted by the present application is: providing a storage medium storing program instructions capable of implementing any of the above dual-tracer ECT imaging methods.

[0040] The beneficial effects of the present application are as follows: the dual-tracer ECT imaging method of the present application obtains first projection data collected by a high-energy ray detector, uses MR images as prior images, and uses a pre-trained PET imaging model to reconstruct images, thereby obtaining high-quality PET images, and then obtains features in the PET image and uses the features in the PET image to guide the second projection data collected by the low-energy ray detector to reconstruct the SPECT image. Compared with the method of reconstructing the image using the second projection data alone, the SPECT image reconstructed by the nuclear feature guidance method using the PET image is of higher quality, and by integrating PET imaging and SPECT imaging, only one projection data acquisition is required, which on the one hand reduces equipment costs, and on the other hand avoids repeated scanning of patients and reduces radiation damage to patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 1 is a schematic flow chart of a dual-tracer ECT imaging method according to an embodiment of the present invention;

[0042] Figure 2 Schematic diagram of the functional modules of a dual-tracer ECT imaging device according to an embodiment of the present invention;

[0043] Figure 3 is a schematic structural diagram of a computer device according to an embodiment of the present invention;

[0044] Figure 4 It is a schematic structural diagram of a storage medium according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0046] The terms "first," "second," and "third" in this application are used only for descriptive purposes and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of such features. In the description of this application, "multiple" means at least two, for example, two, three, etc., unless otherwise specifically defined. All directional indications in the embodiments of this application (such as up, down, left, right, front, back...) are only used to explain the relative positional relationship, movement, etc. between the components under a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications also change accordingly. In addition, the terms "including" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products, or devices.

[0047] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0048] Figure 1 FIG is a flow chart of a dual-tracer ECT imaging method according to an embodiment of the present invention. It should be noted that the method of the present invention is not limited to the dual-tracer ECT imaging method if substantially the same results are achieved. Figure 1 The process sequence shown is limited. Figure 1 As shown, the method includes the steps of:

[0049] Step S101: acquiring first projection data of a target acquired by a high-energy ray detector, second projection data of the target acquired by a low-energy ray detector, and MR image data of the target.

[0050] In this embodiment, the dual-tracer ECT imaging method is applied to a dual-tracer ECT imaging system. The dual-tracer ECT imaging system includes an annular detector. The inner layer of the annular detector is a low-energy radiation detector for emitting low-energy radiation, such as technetium-99m SPECT, which decays to release gamma rays, which emit a 141 keV photon in a random direction. The outer layer of the annular detector is a high-energy radiation detector for emitting high-energy radiation, such as 18F-FDG PET. The positron-emitting nuclide 18F decays to release positrons. These positrons collide and annihilate with negative electrons in the human body within a range of 1-3 mm, simultaneously releasing a pair of photons of equal energy and opposite directions, with a photon energy of 511 keV. It should be noted that the present invention does not limit the external shape of the annular detector; it can be circular or square.

[0051] Step S102: Input the first projection data and the MR image data into a pre-trained PET imaging model, and perform image reconstruction using the MR image data as prior data to obtain a PET image. The PET imaging model is trained based on pre-prepared sample PET projection data and sample MR image data aligned with the sample PET projection data.

[0052] In step S102, after obtaining the first projection data and MR image of the target, the first projection data and MR image are used as input to a pre-trained PET imaging model for image reconstruction to obtain a reconstructed PET image. During the PET image reconstruction, the MR image is introduced as a prior image to assist in image reconstruction, thereby improving the quality of the reconstructed PET image. Specifically, the principle of PET image reconstruction is as follows:

[0053] First, after acquiring the first projection data, we can assume that there is a set of independent random variables, whose expected values ​​are related to the PET image to be reconstructed, and can be expressed by affine transformation as follows:

[0054]

[0055] in, is the expected value, x is the PET image to be reconstructed, P is the probability matrix of the relationship between the nuclide distribution and the detection projection in the PET system, and s is the vector of random scattering events.

[0056] Assuming that the first projection data obtained follows a Poisson distribution, it can be expressed by a log-likelihood function:

[0057]

[0058] Where L(y|x) is the log-likelihood function of the first projection data, y i is the first projection data, and M is the total number of annihilation photon coincidence lines in the PET detector.

[0059] In this embodiment, in order to realize PET image reconstruction using a deep neural network, the PET image to be reconstructed is represented as the following mapping function:

[0060] x=f(θ|z);

[0061] Wherein, f represents the neural network of the PET imaging model, z is the MR image data, and θ is the neural network parameter of the PET imaging model.

[0062] From the above, we can get the integrated formula: Therefore, the above log-likelihood function is expressed in terms of θ as:

[0063]

[0064] At this point, the PET imaging model can be expressed as:

[0065]

[0066] x=f(θ|z);

[0067] The maximum likelihood estimation process of the PET image to be reconstructed is expressed as:

[0068]

[0069]

[0070] in, is the maximum likelihood estimate of the neural network parameters, is the maximum likelihood estimate of the PET image to be reconstructed.

[0071] Due to the coupling between matrix transformation and neural network, the maximum likelihood estimation process of the PET image to be reconstructed is difficult to solve. Therefore, this embodiment converts it into a constraint form:

[0072] maxL(|);

[0073] stx=f(θ|z);

[0074] Furthermore, this embodiment uses an enhanced Lagrangian form to express the above constraints:

[0075]

[0076] Among them, μ and θ are obtained through model training iteration, and ρ is a preset parameter. is the expected value.

[0077] This enhanced Lagrangian form can be solved iteratively using the Alternating Direction Method of Multipliers (ADMM). ADMM decomposes a large global problem into multiple smaller, more easily solvable local subproblems through a decomposition and coordination process, and then obtains the solution to the large global problem by coordinating the solutions of the subproblems.

[0078] In this embodiment, the PET imaging model is implemented based on a 3D U-net network structure. The PET imaging model is pre-trained, and the training process includes:

[0079] 1. Perform a preset number of MLEM reconstructions using the sample PET projection data to obtain a sample reconstructed image.

[0080] 2. Using the sample reconstructed image, label the sample PET projection data.

[0081] 3. Iteratively train the PET imaging model using the labeled sample PET projection data and the sample MR image data registered with the sample PET projection data.

[0082] Specifically, in this embodiment, the sample PET projection data is pre-reconstructed with MLEM for a preset number of times, and the preset number is pre-set, such as 60 times. The sample PET projection data is then labeled using the reconstructed image, and finally the PET imaging model is iteratively trained in combination with the sample MR image data after registration with the sample PET projection data.

[0083] Step S103: constructing a SPECT image by using the PET image to guide the second projection data in a nuclear feature-guided manner.

[0084] Specifically, since the decay of each radioactive atom in the SPECT system releases one photon, rather than two back-pointing photons emitted simultaneously as in the PET system, PET images are often more accurate in locating lesions, i.e., the resolution is higher. Therefore, this embodiment uses PET images to guide SPECT image reconstruction.

[0085] Furthermore, step S103 specifically includes:

[0086] 1. Extract all features of the PET image and construct a kernel matrix based on all features.

[0087] The step of extracting all features of the PET image and constructing a kernel matrix based on all features specifically includes:

[0088] 1.1. Using the radial Gaussian kernel function, all feature data of the PET image are extracted.

[0089] Specifically, the process of extracting the PET image feature data using the radial Gaussian kernel function is expressed as:

[0090]

[0091] Wherein, k is the radial Gaussian kernel function, f j is the jth pixel in the PET image, f l f j One of the neighborhood pixels of , σ is a preset parameter used to adjust the difference between neighborhood pixels.

[0092] 1.2. Based on the nearest neighbor search method, a sparse matrix is ​​established according to all the feature data.

[0093] Specifically, since a complete kernel matrix K is usually large and the corresponding amount of calculation is also large, it is often not directly used in practice. Therefore, this application reduces the amount of calculation by establishing a sparse matrix. It uses the k-nearestneighbors nearest neighbor search method. In the process of establishing the sparse matrix, the distance between pixels is determined by measuring the Euclidean distance between pixels. The specific formula for establishing the sparse matrix is ​​as follows:

[0094]

[0095] Wherein, K is the sparse matrix.

[0096] It should be noted that in order to more easily select the relevant parameters of the matrix, when obtaining the feature data, it is necessary to normalize the feature data, specifically:

[0097]

[0098] in, is the normalized feature data, f j,l is the characteristic data before normalization, σ l (f) is f j The standard deviation of the lth element among all neighborhood pixels.

[0099] 1.3. Regularize the sparse matrix to obtain the kernel matrix.

[0100] Specifically, in order to ensure that the counts are preserved during matrix conversion, the sparse matrix needs to be regularized as follows:

[0101]

[0102] in, is a regularized sparse matrix, diag -1 is the inverse matrix of the diagonal matrix, 1 N is a vector of length N whose elements are all 1, the same size as the rows and columns of the sparse matrix.

[0103] 2. Based on the kernel matrix and the second projection data, a maximum likelihood estimation method is used for iteration to calculate an optimal coefficient image estimation value.

[0104] First, it is important to understand that, assuming that a represents the distribution of radionuclides, i.e., the SPECT image to be reconstructed, and the second projection data b is assumed to be a variable that obeys a Poisson distribution, then the expected value of b is Assuming that it is related to a, the following affine transformation can be obtained:

[0105]

[0106] Where R is the probability matrix of the relationship between the nuclide distribution and the detection projection of the SPECT system, and c is the random scattering event vector.

[0107] Assume that the second projection data b can be modeled as a set of independent Poisson random variables using the log-likelihood function:

[0108]

[0109] Where L(b|a) represents the log-likelihood function of the second projection data b, N is the number of photons captured by the SPECT detector, and b i is the i-th second projection data, is the i-th expected value.

[0110] The maximum likelihood estimate of the SPECT image can be obtained by maximizing the Poisson log-likelihood function:

[0111]

[0112] in, is the maximum likelihood estimate of the SPECT image to be reconstructed.

[0113] At the same time, the EM algorithm can be used to derive the optimal solution of the SPECT image to be reconstructed by the following iterative update steps:

[0114]

[0115] Among them, 1 Nis a vector of length N with all elements being 1, n represents the number of iterations, and T represents the matrix transpose. It should be noted that in the iterative calculation process of the above formula, vector multiplication and division are element-by-element operations.

[0116] In this embodiment, the SPECT image and the kernel matrix are represented by using the coefficient image estimation value, specifically:

[0117] a=Kα;

[0118] Where a is the SPECT image, K is the kernel matrix, and α is the coefficient image estimation value.

[0119] Therefore, in this embodiment, the optimal expression of the SPECT image reconstruction problem can be evolved into the following formula through the above-mentioned nuclear feature-guided method:

[0120]

[0121] in, is the optimal coefficient image estimation value, b is the second projection data, is the kernel matrix, and α is the coefficient image estimation value.

[0122] The expression of the optimal coefficient image estimate is iterated using the maximum EM algorithm combined with Poisson random distribution and is expressed as:

[0123]

[0124] Where R is the probability matrix of the relationship between the nuclide distribution and the detection projection of the SPECT system, c represents the random scattering event vector, T represents the matrix transpose, 1 N is a vector of length N whose elements are all 1, and n is the number of iterations.

[0125] 3. Reconstruct the SPECT image using the optimal numerical image estimation value and the kernel matrix.

[0126] Specifically, after obtaining the optimal coefficient image estimation value, the optimal SPECT image reconstruction process is expressed as:

[0127]

[0128] in, The optimal SPECT image.

[0129] The dual-tracer ECT imaging method of an embodiment of the present invention obtains first projection data collected by a high-energy ray detector, uses MR images as prior images, and reconstructs images using a pre-trained PET imaging model to obtain high-quality PET images. Features in the PET images are then obtained and used to guide second projection data collected by a low-energy ray detector to reconstruct SPECT images. Compared to methods that use only second projection data for image reconstruction, the SPECT image reconstructed using the nuclear feature guidance method using PET images has higher quality. Furthermore, by integrating PET imaging and SPECT imaging, only one projection data acquisition step is required, which reduces equipment costs while avoiding repeated patient scans and reducing radiation damage to patients.

[0130] Figure 2 FIG is a functional module diagram of a dual-tracer ECT imaging device according to an embodiment of the present invention. Figure 2 As shown, the dual-tracer ECT imaging device 20 includes an acquisition module 21 , a PET imaging module 22 and a SPECT imaging module 23 .

[0131] An acquisition module 21 is configured to acquire first projection data of a target acquired by a high-energy ray detector, second projection data of the target acquired by a low-energy ray detector, and MR image data of the target;

[0132] a PET imaging module 22, configured to input the first projection data and the MR image data into a pre-trained PET imaging model, and perform image reconstruction using the MR image data as prior data to obtain a PET image, wherein the PET imaging model is trained based on pre-prepared sample PET projection data and sample MR image data registered with the sample PET projection data;

[0133] The SPECT imaging module 23 is configured to construct a SPECT image by using the PET image to guide the second projection data in a nuclear feature-guided manner.

[0134] Optionally, the PET imaging model is expressed as:

[0135]

[0136] x=f(θ|z);

[0137] Wherein, θ is the neural network parameter of the PET imaging model, L(y|θ) is the log-likelihood function representation of the first projection data, M is the total number of annihilation photon coincidence lines in the PET detector, and y iis the i-th first projection data, x is the PET image to be reconstructed, f represents the neural network of the PET imaging model, and z is the MR image data;

[0138] The maximum likelihood estimation method is used to reconstruct the PET image, which is expressed as:

[0139]

[0140]

[0141] in, is the maximum likelihood estimate of the neural network parameters, is the maximum likelihood estimate of the PET image to be reconstructed.

[0142] Optionally, the SPECT imaging module 23 performs the operation of using the PET image to guide the second projection data to construct a SPECT image in a nuclear feature-guided manner, specifically including: extracting all features of the PET image and constructing a kernel matrix based on all features; iterating based on the kernel matrix and the second projection data using the maximum likelihood estimation method to calculate the optimal coefficient image estimate; and reconstructing the SPECT image using the optimal coefficient image estimate and the kernel matrix.

[0143] Optionally, the expression of the optimal coefficient image estimate is:

[0144]

[0145] in, is the optimal coefficient image estimation value, b is the second projection data, is the kernel matrix, α is the coefficient image estimation value;

[0146] The expression of the optimal coefficient image estimation value is iterated using the maximum EM algorithm combined with Poisson random distribution and is expressed as:

[0147]

[0148] Where R is the probability matrix of the relationship between the nuclide distribution and the detection projection of the SPECT system, c represents the random scattering event vector, T represents the matrix transpose, 1 N is a vector of length N whose elements are all 1, and n is the number of iterations.

[0149] Optionally, the SPECT imaging module 23 performs the operation of extracting all the features of the PET image and constructing a kernel matrix based on all the features, specifically including: extracting all the feature data of the PET image using a radial Gaussian kernel function; establishing a sparse matrix based on all the feature data based on a nearest neighbor search method; and regularizing the sparse matrix to obtain the kernel matrix.

[0150] Optionally, the extraction of the feature data is expressed as:

[0151]

[0152] Wherein, k is the radial Gaussian kernel function, f j is the jth pixel in the PET image, f l f j One of the neighboring pixels of , σ is a preset parameter used to adjust the difference between neighboring pixels;

[0153] The establishment of the sparse matrix is ​​expressed as:

[0154]

[0155] Wherein, K is the sparse matrix;

[0156] The sparse matrix regularization is expressed as:

[0157]

[0158] in, is a regularized sparse matrix, diag -1 is the inverse matrix of the diagonal matrix, 1 N is a vector of length N whose elements are all 1, the same size as the rows and columns of the sparse matrix.

[0159] Optionally, the PET imaging model is implemented based on a 3DU-net network structure; training the PET imaging model includes: performing a preset number of MLEM reconstructions using sample PET projection data to obtain a sample reconstructed image; labeling the sample PET projection data using the sample reconstructed image; and iteratively training the PET imaging model using the labeled sample PET projection data and the sample MR image data aligned with the sample PET projection data.

[0160] For other details of the technical solutions for implementing the modules in the dual-tracer ECT imaging apparatus in the above embodiment, please refer to the description of the dual-tracer ECT imaging method in the above embodiment, which will not be repeated here.

[0161] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similarities between the various embodiments can be referred to in conjunction with each other. For device embodiments, since they are generally similar to method embodiments, their description is relatively simple, and for relevant details, reference can be made to the description of the method embodiments.

[0162] See also Figure 3 , Figure 3 FIG. 1 is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. Figure 3 As shown, the computer device 30 includes a processor 31 and a memory 32 coupled to the processor 31. The memory 32 stores program instructions. When the program instructions are executed by the processor 31, the processor 31 executes the steps of the dual-tracer ECT imaging method described in any of the above embodiments.

[0163] The processor 31 may also be referred to as a CPU (Central Processing Unit). The processor 31 may be an integrated circuit chip having signal processing capabilities. The processor 31 may also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. The general-purpose processor may be a microprocessor or any conventional processor.

[0164] See Figure 4 , Figure 4 Schematic diagram of the structure of the storage medium of an embodiment of the present invention. The storage medium of an embodiment of the present invention stores program instructions 41 that can implement all the above methods, wherein the program instructions 41 can be stored in the above storage medium in the form of a software product, including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or a computer device such as a computer, a server, a mobile phone, or a tablet.

[0165] In the several embodiments provided in this application, it should be understood that the disclosed computer devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0166] In addition, the functional units in the various embodiments of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware or in the form of software functional units. The above is only an implementation method of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the description and drawings of this application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A dual-tracer ECT imaging method, characterized in that: include: Acquiring first projection data of the target acquired by the high-energy ray detector, second projection data of the target acquired by the low-energy ray detector, and MR image data of the target; inputting the first projection data and the MR image data into a pre-trained PET imaging model, and performing image reconstruction using the MR image data as prior data to obtain a PET image, wherein the PET imaging model is trained based on pre-prepared sample PET projection data and sample MR image data registered with the sample PET projection data; constructing a SPECT image by using the PET image to guide the second projection data in a nuclear feature-guided manner; The PET imaging model is expressed as: x=f(θ|z); Wherein, θ is the neural network parameter of the PET imaging model, L(y|θ) is the log-likelihood function representation of the first projection data, M is the total number of annihilation photon coincidence lines in the PET detector, and y i is the i-th first projection data, x is the PET image to be reconstructed, f represents the neural network of the PET imaging model, and z is the MR image data; The maximum likelihood estimation method is used to reconstruct the PET image, which is expressed as: in, is the maximum likelihood estimate of the neural network parameters, is a maximum likelihood estimate of the PET image to be reconstructed; The method of using the PET image to guide the second projection data to construct a SPECT image based on nuclear feature guidance includes: Extracting all features of the PET image and constructing a kernel matrix based on all the features; Iteratively calculating an optimal coefficient image estimation value using a maximum likelihood estimation method based on the kernel matrix and the second projection data; The SPECT image is reconstructed using the optimal numerical image estimate and the kernel matrix.

2. The dual-tracer ECT imaging method according to claim 1, characterized in that: The expression of the optimal coefficient image estimation value is: in, is the optimal coefficient image estimation value, b is the second projection data, is the kernel matrix, α is the coefficient image estimation value; The expression of the optimal coefficient image estimation value is iterated using the maximum EM algorithm combined with Poisson random distribution and is expressed as: Among them, R is the probability matrix of the relationship between the nuclide distribution and the detection projection of the SPECT system, c represents the random scattering event vector, T represents the matrix transpose, 1 N is a vector of length N whose elements are all 1, and n is the number of iterations.

3. The dual-tracer ECT imaging method according to claim 2, characterized in that: The step of extracting all features of the PET image and constructing a kernel matrix based on all features includes: Extracting all feature data of the PET image using a radial Gaussian kernel function; Based on the nearest neighbor search method, a sparse matrix is ​​established according to all the feature data; Regularizing the sparse matrix to obtain the kernel matrix.

4. The dual-tracer ECT imaging method according to claim 3, characterized in that: The extraction of the feature data is expressed as: Wherein, k is the radial Gaussian kernel function, is the jth pixel in the PET image, f l for One of the neighboring pixels of , σ is a preset parameter used to adjust the difference between neighboring pixels; The establishment of the sparse matrix is ​​expressed as: Wherein, K is the sparse matrix; The sparse matrix regularization is expressed as: in, is a regularized sparse matrix, diag -1 is the inverse matrix of the diagonal matrix, 1 N is a vector of length N whose elements are all 1, the same size as the rows and columns of the sparse matrix.

5. The dual-tracer ECT imaging method according to claim 1, wherein: The PET imaging model is implemented based on the 3DU-net network structure; Training the PET imaging model, comprising: Performing a preset number of MLEM reconstructions using the sample PET projection data to obtain a sample reconstructed image; annotating sample PET projection data using the sample reconstructed image; The PET imaging model is iteratively trained using the labeled sample PET projection data and the sample MR image data registered with the sample PET projection data.

6. A dual-tracer ECT imaging device using the dual-tracer ECT imaging method according to claim 1, characterized in that: include: an acquisition module, configured to acquire first projection data of the target acquired by the high-energy ray detector, second projection data of the target acquired by the low-energy ray detector, and MR image data of the target; a PET imaging module, configured to input the first projection data and the MR image data into a pre-trained PET imaging model, and perform image reconstruction using the MR image data as prior data to obtain a PET image, wherein the PET imaging model is trained based on pre-prepared sample PET projection data and sample MR image data registered with the sample PET projection data; A SPECT imaging module is configured to construct a SPECT image by using the PET image to guide the second projection data in a nuclear feature-guided manner.

7. A computer device, characterized in that: The computer device includes a processor and a memory coupled to the processor, wherein program instructions are stored in the memory. When the program instructions are executed by the processor, the processor performs the steps of the dual-tracer ECT imaging method according to any one of claims 1 to 5.

8. A storage medium, characterized in that: The device stores program instructions capable of implementing the dual-tracer ECT imaging method according to any one of claims 1 to 5.

Images