PET image scattering correction method and system

Abstract

The invention relates to a scattering correction method and system for a PET image. The method comprises the following steps: modeling a PET acquisition process; based on an ADMM algorithm, alternately and iteratively solving radioactivity distribution and an attenuation coefficient of a detector space; based on a deep learning network G, mapping from PET detection data and a detector space attenuation coefficient to accurate scattering distribution estimation is realized; keeping the radioactive activity distribution and the spatial attenuation coefficient of the detector as constants, and maximizing a log-like elihold function by using unfitted scattering distribution estimation to obtain a scattering correction factor; and obtaining a PET image subjected to scattering correction by utilizing the accurate scattering distribution after fitting of the spatial attenuation coefficient of the detector and the scattering correction factor. The beneficial effects of the method are that image quality reduction caused by inaccurate scattering evaluation is avoided, scattering correction time consumption is greatly reduced, and stability of high noise data is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of medical imaging technology, and in particular to a scatter correction method and system for PET images. Background Art

[0002] Positron emission tomography (PET) is a cutting-edge nuclear medicine imaging diagnostic technology. In the actual diagnosis and treatment process, this technology uses radioactive isotopes such as 18F and 11C to label metabolites and inject these labeled compounds into the human body. Subsequently, the patient is subjected to functional metabolic imaging through the PET system to vividly display the metabolic activity state in the living body, thereby achieving the goal of disease diagnosis. Currently, commercial PET devices circulating on the market are generally combined with other imaging modes, such as computed tomography (CT) or magnetic resonance imaging (MRI), which can not only synchronously obtain the patient's anatomical structure information, but also significantly improve the accuracy of lesion positioning. Ultimately, the functional and anatomical images are fused on the same device, combining the advantages of the two modalities, allowing doctors to have a more intuitive and comprehensive insight into the patient's physical condition, which helps to detect and accurately diagnose lesions early, especially in the diagnosis and treatment of tumors, heart and brain diseases. Excellent performance has been shown.

[0003] However, during the PET data acquisition process, photons may experience Compton scattering in the human body, causing their direction of travel to change. Given the limitations of the detector's energy resolution, such scattering events are sometimes mistakenly recorded as true coincidence events, which in turn interfere with the position information of the original signal, induce artifacts in the generated image, and seriously affect image quality. Especially in three-dimensional acquisition mode, the proportion of scattered coincidence events may be as high as 30% to 60% of the total counts. Therefore, implementing effective scatter correction has become an indispensable part of the PET reconstruction process.

[0004] The currently commonly used scatter correction technology is the Single Scatter Simulation (SSS) algorithm. This technology estimates the overall scatter distribution by calculating the probability of a single gamma-ray particle undergoing single scattering before detection based on the radioactivity distribution and linear attenuation coefficient distribution of the scanned object. However, since the actual measurement data is mixed with multiple scattering components, the SSS method is difficult to accurately estimate the proportional relationship between the scattering component and the total coincidence rate. Generally speaking, the scattering contribution is determined by analyzing the radial tail characteristics of the sinusoidal graph or by Monte Carlo simulation. In actual clinical acquisition, the accuracy of SSS is usually affected by the following factors:

[0005] First, before scatter correction, the accurate initial distribution of radioactivity is usually unknown, and it is often approximated by using a scatter-free PET reconstructed image. When the patient is heavy, the deviation between the scatter-free PET initial image and the accurate initial distribution of radioactivity is large, so this approximation usually leads to large errors in the scatter distribution estimation. Even if the scattering parameters are iteratively solved in the SSS process, global convergence cannot be guaranteed, and the above errors still exist. The SSS calculation process is complex and time-consuming even with a small number of iterations, which greatly limits the efficiency of scatter correction and increases the time cost of image reconstruction.

[0006] Secondly, for multimodal acquisition, although different imaging technologies can provide us with the necessary attenuation information for scatter estimation, we still face some challenges in actual operation. Taking the PET / CT system as an example, the coverage of PET often exceeds the working area of ​​CT or other auxiliary imaging technologies. Especially for subjects with larger body size, other imaging methods may cause the linear attenuation distribution to be truncated in some areas due to insufficient field of view, thus affecting the quality of PET images. In addition, CT imaging is fast and can be completed almost instantly, while PET takes a long time to complete the scan of a part. This makes it difficult for patients to remain completely still during the entire examination, and factors such as respiratory movement may cause spatial misalignment between PET and CT images. In addition, it is worth noting that if there are metal objects in the subject's body, such as pacemakers, dental implants, etc., obvious "metal artifacts" may appear on the CT image, which also brings additional challenges to subsequent processing. In short, in clinical practice, the problem of incomplete matching of radioactivity distribution images with other modality images due to various reasons will directly affect the effect of scatter correction and may even introduce serious artifacts in the final result. Finally, when applying the tail fitting algorithm of the SSS algorithm, it only relies on the data outside the object as a reference. If the amount of available data is limited, especially for large target objects, this method may become unreliable and easily lead to the phenomenon of incorrect estimation of the scattering distribution. Summary of the invention

[0007] Technical issues to be solved

[0008] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a technical problem of how to further improve the image quality and reduce the reconstruction time of scatter correction of PET images.

[0009] Technical Solution

[0010] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0011] In a first aspect, the present invention provides a scatter correction method for a PET image, comprising:

[0012] Modeling the PET acquisition process;

[0013] Based on the ADMM algorithm, the radioactivity distribution and the attenuation coefficient of the detector space are solved by alternating iteration;

[0014] Based on the deep learning network G, the mapping of PET detection data and detector spatial attenuation coefficient to accurate scattering distribution estimation is realized;

[0015] Keeping the radioactivity distribution and the detector spatial attenuation coefficient constant, the scatter correction factor is obtained by maximizing the log-likelihood function using the unfitted scatter distribution estimate;

[0016] The accurate scatter distribution after fitting the detector spatial attenuation coefficient and scatter correction factor is used to obtain the scatter-corrected PET image.

[0017] Optionally, the PET acquisition process is modeled according to the following formula:

[0018]

[0019] In the formula y=[y 1t ,y 2t , …, y it , …, y NT ]' represents the detected data, represents the average value of the detection data, N represents the size of the detection data sinusoidal graph, T represents the size of the TOF discrete space, i represents the variable index of the detection data sinusoidal graph response line, and t represents the variable index of the TOF discrete space; the single quote superscript represents the matrix transpose operation; x = [x1, x2, ..., x j , …, x M ]' represents the unknown radioactivity distribution image, M represents the size of the radioactivity distribution image space, j represents the variable index of the radioactivity distribution image space; a = [a1, a2, ..., a j …, a N] ' represents the attenuation coefficient of the detector space; A = [A ijt ] is the system matrix; S = [s 1t ,s 2t ,…,s NT ]' represents the estimated value of scattering, α = [α 1t , α 2t , …, α NT ]' represents the correction factor for the scatter estimation at each position.

[0020] Optionally, the radioactivity distribution and the attenuation coefficient of the detector space are solved by alternate iteration based on the ADMM algorithm, including:

[0021] According to the formula The radioactivity distribution x and the detector spatial attenuation coefficient a are solved by alternating iterations.

[0022] Optionally, based on the deep learning network G, a mapping of PET detection data and detector spatial attenuation coefficient to accurate scatter distribution estimation is implemented, including:

[0023] PET detection data x and detector spatial attenuation coefficient a are used as input, and the accurate scatter distribution s label As labels, we train the network and optimize the network parameters θ to minimize the loss function L. Finally, we get the accurate scattering distribution s. The formula is as follows:

[0024]

[0025] Optionally, keeping the radioactivity distribution and the detector spatial attenuation coefficient constant, using the unfitted scattering distribution estimate, maximizing the log-likelihood function, and obtaining the scattering correction factor include:

[0026] Keeping the radioactivity distribution and the attenuation coefficient of the detector space constant, according to the formula Get a new scattering correction factor;

[0027] in,

[0028] Optionally, the training network G includes: a CNN network, a Unet network or a GAN network.

[0029] Optionally, the log-likelihood function is:

[0030]

[0031] In a second aspect, the present invention provides a scatter correction system for PET images, comprising:

[0032] Modeling module, which models the PET acquisition process;

[0033] Iterative module, which solves the radioactivity distribution and the attenuation coefficient of the detector space by alternating iteration based on the ADMM algorithm;

[0034] The mapping module, based on the deep learning network G, realizes the mapping of PET detection data and detector spatial attenuation coefficient to accurate scatter distribution estimation;

[0035] The scatter correction factor determination module keeps the radioactivity distribution and the detector spatial attenuation coefficient as constants, uses the unfitted scatter distribution estimate, maximizes the log-likelihood function, and obtains the scatter correction factor;

[0036] The correction module uses the accurate scattering distribution after fitting the detector spatial attenuation coefficient and the scattering correction factor to obtain a scatter-corrected PET image.

[0037] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed, the method for scatter correction of a PET image as described in any one of the first aspects above is implemented.

[0038] In a fourth aspect, the present invention provides a storage device, comprising a storage medium and a processor, wherein the storage medium stores a computer program, and when the program is executed by the processor, the scatter correction method for a PET image described in any one of the first aspects above is implemented.

[0039] Beneficial Effects

[0040] The beneficial effects of the present invention are as follows: a scatter correction method for PET images of the present invention utilizes an alternating iterative algorithm to obtain the attenuation coefficient of the detector space, and then obtains an accurate scatter estimation distribution through a deep learning network, and finally combines the estimated scatter distribution to perform maximum likelihood estimation iteration on the scatter correction factor to obtain the optimal solution, which not only avoids the image quality degradation caused by inaccurate scatter assessment, but also greatly reduces the time consumption of scatter correction, and the stability to high-noise data is also greatly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic flow chart of a scatter correction method for PET images provided in an embodiment of the present invention;

[0042] Figure 2 A PET reconstructed image obtained by calculating scattering using a traditional algorithm provided in an embodiment of the present invention;

[0043] Figure 3 The embodiment of the present invention provides a PET reconstructed image obtained by calculating the scattering distribution using the present invention. DETAILED DESCRIPTION

[0044] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0045] The present invention directly obtains the attenuation coefficient of the detector space from the PET data, and then uses a deep learning network to obtain an accurate scattering distribution. Finally, the optimal scattering correction factor is iteratively solved through the PET detection data to obtain a numerically accurate scattering distribution through fitting. This not only greatly reduces the time cost of scattering correction, but also greatly improves the accuracy of scattering correction.

[0046] First, since the SSS estimation algorithm requires accurate radioactivity distribution and linear attenuation coefficient distribution for model estimation, the initial condition lacks accurate radioactivity distribution information, and the final scattering estimation can only be obtained through multiple iterations based on the PET image without scattering correction, so the error is large and it is relatively time-consuming. This patent directly obtains accurate scattering distribution through a deep learning network, which reduces the dependence on the initial radioactivity and linear attenuation coefficient and greatly shortens the calculation time of scattering correction.

[0047] Secondly, extracting attenuation information directly from PET acquisition data without relying on other modal imaging can effectively eliminate attenuation artifacts caused by mismatching data from different modalities, thereby broadening the application range of PET imaging. By applying PET acquisition data to obtain the attenuation coefficient of the detector space, it is ensured that the PET distribution and attenuation distribution are strictly matched, effectively eliminating motion artifacts and improving image quality.

[0048] Finally, in the process of fitting the scatter correction factor, this method fully utilizes all data for evaluation through iterative calculations and accurately models the scattering process. Compared with the traditional tail fitting method, this method has a higher tolerance for noise in the acquired data and the ability to handle redundancy. This patent only uses PET data to more accurately estimate the scatter component, thereby optimizing the PET reconstruction process. This method not only enhances the applicability and accuracy for patients with large body weight, but also achieves independence from other modality images without dependence. At the same time, this method also improves the robustness of scatter correction to data noise, ensuring that more accurate PET images can be obtained in clinical acquisitions, effectively eliminating scatter correction artifacts.

[0049] In order to better understand the above technical solution, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0050] First, refer to Figure 1 This embodiment provides a scatter correction method for a PET image, comprising:

[0051] S1, Modeling the PET acquisition process.

[0052] S2, based on the ADMM algorithm, alternately iterates to solve the radioactivity distribution and the attenuation coefficient in the detector space.

[0053] S3, based on the deep learning network G, realizes the mapping of PET detection data and detector spatial attenuation coefficient to accurate scattering distribution estimation.

[0054] S4, keeping the radioactivity distribution and the detector spatial attenuation coefficient constant, using the unfitted scattering distribution estimate to maximize the log-likelihood function and obtain the scattering correction factor.

[0055] S5, using the accurate scattering distribution fitted by the detector spatial attenuation coefficient and the scattering correction factor to obtain a scattering-corrected PET image.

[0056] Optionally, the PET acquisition process is modeled according to the following formula:

[0057]

[0058] In the formula y=[y 1t ,y 2t , …, y it , …, y NT ]' represents the detected data, represents the average value of the detection data, N represents the size of the detection data sinusoidal graph, T represents the size of the TOF discrete space, i represents the variable index of the detection data sinusoidal response line LOR (line of response), and t represents the variable index of the TOF discrete space. The single quote superscript represents the matrix transposition operation. x = [x1, x2, ..., x j , …, x M ]' represents the unknown radioactivity distribution image, M represents the size of the radioactivity distribution image space, j represents the variable index of the radioactivity distribution image space, and represents the point source at the corresponding spatial position. a=[a1,a2,…,a j …, a N] ' represents the attenuation coefficient of the detector space. A=[A ijt ] is the system matrix, which expresses the probability that a point source j in the PET system is detected by a response line LOR i and has a flight time TOF of t in mathematical form, reflecting the physical characteristics of the system. 1t ,s 2t ,…,s NT ]' represents the estimated value of scattering, α = [α 1t , α 2t , …, α NT ]' represents the correction factor of the scatter estimation at each position, which is used to ensure that the estimated scatter component is consistent with the scatter component of the actual collected data.

[0059] PET detection data follows Poisson distribution, and the unknowns are the radioactivity distribution x, the attenuation coefficient a in the detector space, and the scatter correction factor α. The log-likelihood function of the detection data is expressed as:

[0060]

[0061] Substituting formula (1) into formula (2) and ignoring terms unrelated to the unknown, the log-likelihood function can be written as:

[0062]

[0063] Using PET detection data to estimate the attenuation coefficient a in the detector space, ignoring the influence of the scattered component, formula (3) becomes:

[0064]

[0065] Optionally, the radioactivity distribution and the attenuation coefficient of the detector space are solved by alternate iteration based on the ADMM algorithm, including:

[0066] According to the formula The radioactivity distribution x and the detector spatial attenuation coefficient a are solved by alternating iterations.

[0067] The alternating direction method of multipliers (ADMM) algorithm is used to iteratively solve the radioactivity distribution x and the attenuation coefficient a in the detector space.

[0068] First, ensure that the attenuation coefficient a in the detector space is constant and maximize the log-likelihood function, which is the traditional MLEM algorithm:

[0069]

[0070] In the formula, n represents the current iteration number.

[0071] The initial value of x in formula (5) can be set to a constant or an image obtained by other reconstruction methods or an uncorrected PET image. In this application, it is set to a constant distribution within the imaging field of view, and the constant is selected as 1000.

[0072] Then ensure that the radioactivity distribution x is a constant, maximize the log-likelihood function for the attenuation coefficient a of the unknown detector space, and take the derivative of the attenuation coefficient a of the detector space, that is, formula (6):

[0073]

[0074] Keep all variables non-negative and monotonic:

[0075]

[0076] Substituting formula (6) into formula (7), we get:

[0077]

[0078] In the formula, n represents the current iteration number.

[0079] The initial value of a in formula (8) is usually set to a constant. Since the evaluation algorithm in the formula is only related to the radioactivity distribution, the detector spatial attenuation coefficient obtained by this method strictly matches the radioactivity distribution.

[0080] In order to reduce the calculation time of scatter correction and increase the accuracy of scatter correction, this application applies a deep learning network to obtain accurate scatter distribution estimation, and uses the network G to realize the mapping of PET detection data x and detector spatial attenuation coefficient a to accurate scatter distribution s.

[0081] Optionally, based on the deep learning network G, a mapping of PET detection data and detector spatial attenuation coefficient to accurate scatter distribution estimation is implemented, including:

[0082] PET detection data x and detector spatial attenuation coefficient a are used as input, and the accurate scatter distribution s label As labels, we train the network and optimize the network parameters θ to minimize the loss function L. Finally, we get the accurate scattering distribution s. The formula is as follows:

[0083]

[0084] s label Represents the scattering distribution that meets the conditions for training the network, which comes from the screened training data set. The training data set can come from simulation or actual acquisition. The actual acquired training data set needs to be preprocessed and screened to ensure that the scattering distribution is accurate. In particular, for target objects with large volumes, the data should not be inaccurate due to too few scattering calculation iterations. At the same time, it should avoid the situation where the training data scattering is inaccurate due to mismatched attenuation correction or inaccurate initial scattering activity distribution. Since the tail fitting of the scattering component is easily affected by noise, it will cause the network parameters to overfit, so the network label s label No fitting is performed and it is directly obtained through the SSS algorithm, which improves the robustness of the network training process.

[0085] The training network G can select CNN network, Unet network, GAN network or other networks.

[0086] Optionally, keeping the radioactivity distribution and the detector spatial attenuation coefficient constant, using the unfitted scattering distribution estimate, maximizing the log-likelihood function, and obtaining the scattering correction factor include:

[0087] Keeping the radioactivity distribution and the attenuation coefficient of the detector space constant, according to the formula Get a new scattering correction factor;

[0088] in,

[0089] In order to determine the proportional relationship between the unfitted scattered component and the actual scattered component, it is necessary to fit it, that is, to solve for the appropriate α.

[0090] Formula (1) is written in matrix form as shown in formula (10):

[0091] Y=A×x+S×α(10)

[0092] A is the system matrix, S is the diagonal matrix It can be seen from formula (10) that x and α are symmetrical in the formula. Therefore, the iterative algorithm of the unknown x can be used to solve the unknown α. Therefore, formula (10) can be simplified to the following formula (11), that is, the radioactivity distribution x is kept, the attenuation coefficient a in the detector space is constant, the log-likelihood function is maximized for the unknown scatter correction factor α, and the new scatter correction factor α is directly calculated using PET data. The corresponding algorithm is:

[0093]

[0094] in

[0095] The initial value of α in formula (11) can be set to a constant or a scattering correction factor distribution obtained by other fitting methods. In this embodiment, it is set to a constant of 1.

[0096] In the implementation process of the present invention, first, in order to obtain the attenuation coefficient a of the detector space, the objective function is maximized for the radioactive activity distribution x, that is, the traditional MLEM iterative reconstruction algorithm, and then the radioactive activity distribution x is kept as a constant, and the objective function is maximized for the attenuation coefficient a of the unknown detector space to obtain the attenuation coefficient a of the detector space; secondly, a deep learning network is applied to combine the radioactive activity distribution x and the attenuation coefficient a of the detector space to obtain the unfitted scattering distribution s; finally, the radioactive activity distribution x and the attenuation coefficient a of the detector space are kept as constants, and the objective function is maximized for the scattering correction factor α, and the estimated unfitted scattering distribution is used for iterative calculation, and the scattering correction factor is continuously corrected to make it close to the actual situation, so as to obtain the scattering correction factor that meets the requirements of the maximized objective function, and finally the matched detector space attenuation coefficient and the accurate scattering estimate after fitting are applied to the reconstruction, which not only improves the accuracy of the scattering correction, but also greatly reduces the reconstruction time.

[0097] Figure 2 For the PET reconstructed image of a uniform barrel source with a radius close to the edge of the field of view obtained by calculating the scatter using the traditional algorithm, the center of the barrel source has obviously high uptake and the scatter assessment is inaccurate; Figure 3In order to calculate the scattering distribution using the patented method, the PET images obtained under the same reconstruction method and parameters have significantly improved the unevenness of the barrel source, the image is more uniform, the scattering distribution is more accurate, and the image quality is better.

[0098] This embodiment proposes a more accurate scatter correction method for PET images, which extracts the detector spatial attenuation coefficient from the PET acquisition data and derives the accurate scatter distribution with the help of a deep learning network. The primary advantage is that this method not only effectively overcomes the scatter estimation bias caused by the inaccurate initial activity distribution, but also avoids the significant time consumption caused by the complex iterative calculation of SSS. Secondly, this method realizes the independence of other modal images, and can ensure the accurate acquisition of scatter distribution even in the absence of other modal image support or the poor quality of these images. Furthermore, the scatter correction factor is calculated by an iterative algorithm, which solves the problem of limited data volume and limited stability when the patient is heavy. In addition, while significantly reducing the radiation dose received by the patient, this solution improves the accuracy and stability of the scatter distribution estimation, thereby enhancing its practical value. Compared with the traditional scatter correction algorithm, the present application not only has higher correction accuracy, but also takes less time, which is of great benefit to improving image quality and shortening reconstruction time.

[0099] The following is a specific implementation case of clinical acquisition data to further illustrate a scatter correction method for PET images of the present invention:

[0100] The patient was scanned with a PET device. The PET images were reconstructed using the ordered subsets expectation-maximization (OSEM) method with 17 subsets, 2 iterations, an image matrix of 192x192, and a Gaussian filter half-height width of 4 mm.

[0101] Deep learning network model parameters: network model structure of feature pyramid, loss function uses a mixed loss function of MS-SSIM loss and perceptual difference loss. The training rounds are 500 rounds, with a batch size of 128, and the Adam optimization operator is used to train the network. The learning rate decay value is set to 0.0001, and the rest of the parameters use the default values. The first five rounds use the warm up strategy, and the learning rate increases linearly from 0.0001 to 0.0005. In the remaining training process, the cosine decay learning rate scheduler is used to gradually decay the learning rate to 0.

[0102] x is set to the default initial value of 1000, a is set to the initial value of 1, α is set to the initial value of 1, the number of iterations of the attenuation coefficient in the detector space is 3, the number of iterations of the scatter correction factor is 3, and the reconstruction results are shown in Figure 2 and Figure 3 . Figure 2The PET reconstructed image of a uniform barrel source with a radius close to the edge of the field of view shows that the center of the barrel source has obviously high uptake and inaccurate scatter assessment; Figure 3 In order to calculate the scattering distribution by using the method of the present invention, the non-uniformity phenomenon of the image of the larger volume barrel source is significantly improved in the PET image obtained under the same reconstruction method and parameters.

[0103] In a second aspect, the present embodiment provides a scatter correction system for PET images, comprising: a modeling module for modeling the PET acquisition process; an iteration module for alternately iterating and solving the radioactivity distribution and the attenuation coefficient of the detector space based on the ADMM algorithm; a mapping module for mapping the PET detection data and the detector space attenuation coefficient to an accurate scatter distribution estimate based on a deep learning network G; a scatter correction factor determination module for keeping the radioactivity distribution and the detector space attenuation coefficient as constants, and using an unfitted scatter distribution estimate to maximize the log-likelihood function to obtain a scatter correction factor; and a correction module for obtaining a scatter-corrected PET image using the accurate scatter distribution fitted with the detector space attenuation coefficient and the scatter correction factor.

[0104] According to the scatter correction system for PET images provided in this embodiment, since it is used to implement the steps of a scatter correction method for PET images provided in the first aspect of the present invention, the scatter correction system for PET images has all the technical effects of the scatter correction method for PET images, which will not be repeated here.

[0105] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed, the method for scatter correction of a PET image as described in any one of the first aspects above is implemented.

[0106] In a fourth aspect, an embodiment of the present invention provides a storage device, including a storage medium and a processor, wherein the storage medium stores a computer program, and when the program is executed by the processor, the scatter correction method for a PET image described in any one of the first aspects above is implemented.

[0107] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention should also include these modifications and variations.

[0109] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may alter, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A scatter correction method for PET images, characterized in that: include: Modeling the PET acquisition process; Based on the ADMM algorithm, the radioactivity distribution and the attenuation coefficient of the detector space are solved by alternating iteration; Based on the deep learning network G, the mapping of PET detection data and detector spatial attenuation coefficient to accurate scattering distribution estimation is realized; Keeping the radioactivity distribution and the detector spatial attenuation coefficient constant, the scatter correction factor is obtained by maximizing the log-likelihood function using the unfitted scatter distribution estimate; The accurate scatter distribution after fitting the detector spatial attenuation coefficient and scatter correction factor is used to obtain the scatter-corrected PET image.

2. A scatter correction method for PET images according to claim 1, characterized in that: The PET acquisition process is modeled according to the following formula: In the formula y=[y 1t ,y 2t , …, y it , …, y NT ]' represents the detected data, represents the average value of the detection data, N represents the size of the detection data sinusoidal graph, T represents the size of the TOF discrete space, i represents the variable index of the detection data sinusoidal graph response line, and t represents the variable index of the TOF discrete space; the single quote superscript represents the matrix transpose operation; x = [x1, x2, ..., x j , …, x M ]' represents the unknown radioactivity distribution image, M represents the size of the radioactivity distribution image space, j represents the variable index of the radioactivity distribution image space; a = [a1, a2, ..., a j …, a N] ' represents the attenuation coefficient of the detector space; A = [A ijt ] is the system matrix; S = [s 1t ,s 2t ,…,s NT ]' represents the estimated value of scattering, α = [α 1t , α 2t , …, α NT ]' represents the correction factor for the scatter estimation at each position.

3. The scatter correction method for PET images according to claim 2, characterized in that: Based on the ADMM algorithm, the radioactivity distribution and the attenuation coefficient of the detector space are solved by alternating iteration, including: According to the formula The radioactivity distribution x and the detector spatial attenuation coefficient a are solved by alternating iterations.

4. The scatter correction method for PET images according to claim 3, characterized in that: Based on the deep learning network G, the mapping of PET detection data and detector spatial attenuation coefficient to accurate scatter distribution estimation is realized, including: PET detection data x and detector spatial attenuation coefficient a are used as input, and the accurate scatter distribution s label As labels, we train the network and optimize the network parameters θ to minimize the loss function L. Finally, we get the accurate scattering distribution s. The formula is as follows:

5. The scatter correction method for PET images according to claim 4, characterized in that: Keeping the radioactivity distribution and the detector spatial attenuation coefficient constant, the unfitted scattering distribution is estimated and the log-likelihood function is maximized to obtain the scattering correction factor, including: Keeping the radioactivity distribution and the attenuation coefficient of the detector space constant, according to the formula Get a new scatter correction factor; in, 6. The scatter correction method for PET images according to claim 5, characterized in that: The training network G includes: CNN network, Unet network or GAN network.

7. The scatter correction method for PET images according to claim 6, characterized in that: The log-likelihood function is:

8. A scatter correction system for PET images, characterized in that: include: Modeling module, which models the PET acquisition process; Iterative module, which solves the radioactivity distribution and the attenuation coefficient of the detector space by alternating iteration based on the ADMM algorithm; The mapping module, based on the deep learning network G, realizes the mapping of PET detection data and detector spatial attenuation coefficient to accurate scatter distribution estimation; The scatter correction factor determination module keeps the radioactivity distribution and the detector spatial attenuation coefficient as constants, uses the unfitted scatter distribution estimate, maximizes the log-likelihood function, and obtains the scatter correction factor; The correction module uses the accurate scattering distribution after fitting the detector spatial attenuation coefficient and the scattering correction factor to obtain a scatter-corrected PET image.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the scatter correction method for a PET image as claimed in any one of claims 1 to 7 is implemented.

10. A storage device comprising a storage medium and a processor, wherein the storage medium stores a computer program, wherein: When the processor executes the computer program, the scatter correction method for a PET image according to any one of claims 1 to 7 is implemented.

Images