Method for determining brain PET standardized reconstruction parameters based on clinical brain MRI and PET data

Abstract

The invention discloses a method for determining brain PET standardized reconstruction parameters based on clinical brain MRI and PET data, and relates to the field of medical image processing. The method comprises the following steps: firstly, collecting clinically paired MRI (Magnetic Resonance Imaging) and PET (Positron Emission Tomography) images for preprocessing, carrying out virtual reconstruction by combining a partial volume correction algorithm and system parameters of equipment to estimate'real 'brain activity distribution so as to obtain a first simulated PET image, and calculating difference or similarity between the first simulated PET image and a clinical PET image; through multiple iterations, obtaining an optimal'real 'brain activity distribution diagram, and processing the optimal'real' brain activity distribution diagram into a'standardized PET image 'according to standardized Gaussian filtering; then, constructing different candidate combinations of reconstruction parameters, inputting an optimal'real 'brain activity distribution diagram, traversing a second simulated PET image generated by simulation reconstruction under each combination, and respectively performing difference or similarity calculation with the'standardized PET image', so as to obtain an optimal'real 'brain activity distribution diagram; and selecting the group of candidate reconstruction parameters with the minimum difference or the highest similarity as final standardized reconstruction parameters. The method gets rid of dependence on a physical motif.

Description

Technical Field

[0001] The present invention relates to the field of medical image processing, and in particular to a method for determining brain PET standardized reconstruction parameters based on clinical brain MRI and PET data. Background Art

[0002] Brain PET imaging is an indispensable imaging technique for the diagnosis, staging, treatment efficacy assessment, and brain function research of neurological diseases (such as Alzheimer's disease, Parkinson's disease, epilepsy, and brain tumors). It reveals physiological and biochemical information such as brain tissue metabolism, blood flow, and neurotransmitter receptors by detecting the distribution of radioactive tracers within the brain.

[0003] However, different PET scanning devices (from different manufacturers or different models from the same manufacturer), as well as different reconstruction algorithms and parameter settings, can lead to significant differences in spatial resolution, noise levels, and quantitative accuracy in the resulting PET images. This variability severely hinders direct comparison and data aggregation of PET images across different medical institutions and time points, limiting the development of large-scale multicenter clinical studies, quantitative clinical evaluations, and the development of standardized clinical guidelines.

[0004] like Figure 1 As shown: In multicenter PET imaging, even when faced with the same "real" radioactivity distribution (the brain image on the far left of the figure), after data acquisition and image reconstruction using PET systems at different sites (or equipment) (illustrated here as "Site A PET acquisition and reconstruction" and "Site B PET acquisition and reconstruction"), the final PET images (the upper and lower brain images in the middle of the figure) will have significant differences in visual characteristics (such as clarity and noise level).

[0005] Results vary with different reconstruction parameters at the same site: The two curves (black and orange) for Site A demonstrate that, even on the same device, changing reconstruction parameters (for example, changing the number of subsets from 5 to 10) results in different image resolution characteristics as a function of filtering. Results also vary with the same nominal reconstruction parameters at different sites: Comparing the orange curve (i10s10) for Site A and the blue curve (i10s10) for Site B, even with the same number of iterations and subsets, the image resolution characteristics obtained at the two sites differ. Achieving the target resolution (pink area) may require different post-processing filter FWHM values.

[0006] To achieve the harmonization or standardization of brain PET images, i.e. to make PET images from different sources comparable, a key step is to determine a set of "standardized" reconstruction parameters suitable for a specific device and scanning protocol. Traditionally, this process usually relies on physical phantoms, such as the Hoffman 3D brain phantom. The operator needs to follow a specific procedure to accurately fill the phantom with radionuclide, simulating the activity distribution in human brain, then perform scanning and image reconstruction. By analyzing the quality of the phantom image (e.g. resolution, uniformity, contrast recovery coefficient, etc.) and comparing with the target standard, the optimal reconstruction parameters are adjusted and selected.

[0007] Although the Hoffman phantom method is a recognized reference method, it has many inconveniences in actual clinical operation:

[0008] • Complex operation: The steps of accurate filling of the phantom, avoiding air bubbles, activity calibration, etc. are cumbersome and time-consuming.

[0009] • High cost: Requires the purchase and maintenance of a dedicated phantom, and each use involves radioactive material handling.

[0010] • Difficult to promote clinically: Difficult to be implemented frequently as a routine quality control or parameter optimization method in busy clinical departments.

[0011] • Limited representation: The phantom is an idealized model, which is different from the complex anatomical structure and physiological variation of the real patient brain.

[0012] Therefore, there is an urgent need for a more convenient and efficient method that can fully utilize clinical real data to determine the standardized reconstruction parameters of brain PET, to get rid of the dependence on physical phantoms and promote the wide standardization application of brain PET technology.

[0013] At the same time, the paired image data obtained clinically is diverse, which may come from integrated PET / MRI devices, or from independent MRI scanning after PET / CT scanning, or from PET and MRI scanning at different time points; an ideal method of determining standardized parameters should be able to adapt to different data acquisition scenarios. SUMMARY

[0014] In order to overcome the problems of complex operation, high cost and limited clinical promotion in the prior art when determining brain PET standardization reconstruction parameters by using a physical phantom (such as a Hoffman phantom), the present application provides a method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data, which uses PET data actually scanned in a clinic and paired MRI data, determines standardization reconstruction parameters suitable for a specific PET device through specific image processing and virtual simulation processes, and can adapt to paired PET and MRI data of different sources,

[0015] The method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data specifically comprises the following steps:

[0016] Step one, given initial reconstruction parameters, collect clinical PET scan images and paired T1W-MRI images of a patient, and perform brain structure segmentation on the T1W-MRI images to obtain MRI segmentation results.

[0017] The reconstruction parameters include an iterative algorithm type, an iteration number, a subset number, a filter type, a cutoff frequency and the like;

[0018] The brain structure segmentation refers to accurately segmenting main brain tissue regions such as gray matter (GM), white matter (WM) and cerebrospinal fluid (CSF), or more detailed brain atlas defined brain regions.

[0019] Step two, correct the partial volume effect of the registered clinical PET scan images by using the MRI segmentation results through a partial volume correction algorithm.

[0020] Step three, map the activity value of the PET scan images after the partial volume correction processing to a high-resolution MRI space grid according to the brain regions defined by the MRI segmentation, as an initial estimation of the “real” brain activity distribution.

[0021] Step four, based on the current “real” brain activity distribution map, combine the initial reconstruction parameters and the system parameters of the PET device, and perform a simulation reconstruction process of the PET image to generate a first simulated PET image.

[0022] The system parameters of the PET device include system geometry, detector response function / point spread function model, attenuation map, scatter correction model and the like;

[0023] Step five, compare the first simulated PET image with the real clinical PET scan image, calculate the difference or similarity between the two, and adjust the activity value in the “real” brain activity distribution map according to the difference by using an optimization algorithm.

[0024] The difference or similarity algorithm includes: mean square error, normalized mutual information, structural similarity index, etc. as the objective function.

[0025] The optimization algorithm includes: gradient descent method, conjugate gradient method, algorithm under the expectation maximization (EM) framework, etc.

[0026] The adjustment aims to minimize the difference between the simulated PET image generated by the next virtual scan and the real clinical PET image.

[0027] Step six, return to step four, repeat the simulation reconstruction, and perform multiple iterations until the difference between the simulated PET image and the real PET image is less than the preset convergence threshold, or the maximum number of iterations is reached, and the optimal "real" brain activity distribution map is obtained.

[0028] Step seven, convert the optimal "real" brain activity distribution map into a "standardized PET image" by applying Gaussian filtering processing (standardization processing - i.e. effective spatial resolution);

[0029] The standardization requirement usually refers to the spatial resolution of the image should reach a certain recognized standard; for example, the resolution recommended by the European Nuclear Medicine Society EARL standard, or the uniform resolution defined in other multi-center research protocols.

[0030] Step eight, construct different candidate combinations of reconstruction parameters, and use the optimal "real" brain activity distribution map as the input source image, traverse each candidate reconstruction parameter combination, combine the PET device system parameters, and perform simulation reconstruction to generate the corresponding second simulated PET image for each group;

[0031] Step nine, compare each corresponding second simulated PET image with the "standardized PET image", calculate the difference or similarity between the two, and select the candidate reconstruction parameter group with the smallest difference or the highest similarity as the standardized reconstruction parameter for the PET device for this type of clinical brain scan.

[0032] The advantages of the present application are:

[0033] 1. A method for determining brain PET standardized reconstruction parameters based on clinical brain MRI and PET data, without the need for a physical phantom, completely eliminating the dependence on Hoffman's physical phantom, significantly simplifying the operation process, reducing the implementation cost and human resource demand, and making it possible to determine or verify standardized parameters in the conventional clinical workflow.

[0034] 2. A method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data, which directly uses the actual scan data (including PET and paired MRI) of patients using clinical real data, so that the finally determined reconstruction parameters are closer to the actual application scene in the clinic, and may be more clinically applicable and representative than the parameters obtained based on idealized phantoms.

[0035] 3. A method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data, which has high flexibility and can adapt to paired PET and MRI data from different sources, whether it is PET / MRI integrated machine data or PET / CT combined with independent MRI data.

[0036] 4. A method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data, which has great automation potential, and the entire processing flow from data input to parameter output has high automation potential and can be realized through software integration, improving work efficiency and the objectivity of results.

[0037] 5. A method for determining brain PET standardization reconstruction parameters based on clinical data, which promotes multi-center research and clinical standardization: provides a convenient and effective technical means for different medical institutions and different PET devices to realize the standardization and synergy of brain PET images, helps to improve the data quality and reliability of multi-center clinical research, and promotes the establishment of standardized clinical practice guidelines. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 The present application collects data and reconstructs PET images for PET systems in different existing technologies;

[0039] Figure 2 The present application is a schematic diagram of a method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data;

[0040] Figure 3 The present application is a flowchart of a method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data. DETAILED DESCRIPTION

[0041] The embodiments of the present application will be described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0042] The application discloses a method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data, which does not need a physical phantom, utilizes patient clinical brain positron emission tomography (PET) data and matched magnetic resonance imaging (MRI) data, and determines PET image standardization reconstruction parameters through a specific processing procedure, and is suitable for promoting the cooperation and standardization of brain PET imaging in multicenter research and clinical application.

[0043] As shown in the figure, the method mainly comprises three steps: PET and matched MRI preprocessing, PET virtual scanning and reconstruction, and solving PET reconstruction parameters meeting standardization requirements. Figure 2 Specifically, firstly, the clinical PET scanning data of the patient and the matched high-resolution T1 weighted MRI (T1W-MRI) image are utilized, and the MRI segmentation result is taken as a prior anatomical restriction condition of brain structure and activity distribution. Secondly, an approximate "real" high-resolution brain radioactivity distribution is inversely calculated from the clinical PET data by combining a partial volume correction algorithm and iterative virtual reconstruction (or virtual scanning) based on device system parameters. Finally, the "real" activity distribution is taken as a basis, and the reconstruction parameters are traversed and optimized by performing parameterized virtual scanning again and comparing with a predefined standardization brain PET image target, so that a set of reconstruction parameters capable of making the reconstruction result of the device closest to the standardization target is determined.

[0044] The method for determining brain PET standardization reconstruction parameters based on clinical brain MRI and PET data, as shown in the figure, comprises the following specific steps: Figure 3

[0045] Step one, given initial reconstruction parameters, collect the clinical PET scanning image of the patient and the matched T1W-MRI image, and perform brain structure segmentation on the T1W-MRI image to obtain the MRI segmentation result.

[0046] The reconstruction parameters include iterative algorithm type, iteration number, subset number, filter type, cutoff frequency and the like;

[0047] The specific steps are as follows:

[0048] 1.1, obtain the clinical brain PET image data of the patient, including the image reconstructed by using a conventional acquisition protocol and clinical commonly used (or default) reconstruction parameters of a current PET device (for example, a PET / CT or PET / MRI scanner).

[0049] 1.2, obtain the high-resolution T1 weighted MRI (T1W-MRI) image of the patient matched with the PET image, which is derived from:

[0050] In-machine acquisition with the PET (for example, on a PET / MRI device); ​

[0051] PET is acquired at a different time point but can be aligned with PET images through reliable image registration techniques.

[0052] PET is acquired at a different time point but can be aligned with PET images through reliable image registration techniques.

[0053] 1.3. Accurately register the acquired clinical PET image to its paired T1W-MRI image space.

[0054] 1.4. Perform brain structure segmentation using the T1W-MRI image: accurately segment the major brain tissue regions such as gray matter (GM), white matter (WM), cerebrospinal fluid (CSF), or more detailed brain atlas defined regions.

[0055] MRI segmentation can employ tools or algorithms provided by mature brain image analysis software packages (e.g. FAST of FSL, SPM, FreeSurfer, etc.); PET and MRI registration can employ rigid or affine registration algorithms based on mutual information, normalized mutual information, etc. measures; importantly, ensure the spatial consistency of PET and MRI data. These segmentation results will serve as anatomical prior constraints for activity distribution estimation and partial volume effect correction in subsequent steps.

[0056] Step two, perform partial volume effect correction on the registered clinical PET scan image using the MRI segmentation results through a partial volume correction algorithm (PVC).

[0057] The application of partial volume correction is one of the key steps, the purpose of which is to use the high-resolution anatomical structure information provided by MRI to improve the quantitative accuracy of PET images, to provide more reliable initial activity estimates for subsequent iterations, and to serve as anatomical constraints.

[0058] Using the MRI segmentation results (i.e. anatomical constraints), perform partial volume effect correction on the registered clinical PET image, which aims to compensate for the diffusion and mixing effects of signals between different tissues due to the limited spatial resolution of the PET system, so as to more accurately estimate the true average activity in each brain region.

[0059] The algorithms that can be used for brain PET partial volume correction include but are not limited to:

[0060] The method based on geometric transfer matrix (GTM) proposed by Meltzer et al. (Meltzer CC, et al. J Nucl Med. 1990;21(4):679-88);

[0061] Two-parameter (MG) or three-parameter method proposed by Muller-Gartner et al. (Muller-Gartner HW, et al. J Cereb Blood Flow Metab. 1992;12(4):571-83);

[0062] Region-based iterative deconvolution method proposed by Rousset et al. (Rousset OG, et al. J Nucl Med. 1998;39(5):904-11);

[0063] Van Cittert iterative deconvolution method and its improved version for PET;

[0064] Correction methods based on accurate modeling of point spread function (PSF), such as variants of Richardson-Lucy deconvolution algorithm;

[0065] Machine learning or deep learning driven methods for PVC using MRI priors.

[0066] Step three, map the activity values of the PET scan image after partial volume correction to the high-resolution MRI spatial grid according to the brain regions defined by MRI segmentation, as the initial estimate of the “true” brain activity distribution.

[0067] From the clinical PET image affected by partial volume effect and limited resolution, combined with high-resolution MRI anatomical information, a high-resolution brain radioactivity distribution map as close as possible to the physiological reality is inversed.

[0068] Step four, based on the current “true” brain activity distribution map, combined with the actual reconstruction parameters used when the original PET image was acquired and the system parameters of the PET device, a simulation reconstruction process of the PET image is performed to generate the first simulated PET image.

[0069] The system parameters of the PET device include the detector response function / point spread function (PSF), system geometric configuration, generation method of attenuation map or actual attenuation data, scatter correction model and parameters, etc.

[0070] The virtual scanning (or simulated reconstruction) process is a PET image simulated reconstruction process based on the approximate "real" brain activity distribution, the reconstruction parameters to be evaluated, and the system parameter model of the PET device; it is also a simulation of the physical imaging process of the PET scanner, and the core is the forward projection operation, that is, a given activity distribution map is projected according to the system model (including geometric, blur, attenuation, scattering, etc. Effect) to obtain simulated detector data (sinogram), and then the simulated image is obtained through the reconstruction algorithm (usually consistent or similar to the actual reconstruction algorithm used by the device); An accurate device system matrix or point spread function model is the key to ensuring the simulation accuracy.

[0071] Step five, compare the first simulated PET image with the real clinical PET scan image, calculate the difference or similarity between the two, and adjust the activity value in the "real" brain activity distribution according to the difference, and use the optimization algorithm to adjust the activity value in the "real" brain activity distribution;

[0072] The difference or similarity calculation algorithm includes: mean square error, normalized mean square error, mutual information, structural similarity index, etc. as the objective function;

[0073] Optimization algorithms such as gradient descent method, conjugate gradient method, algorithms under the expectation maximization (EM) framework, etc.

[0074] The goal of adjustment is to minimize the difference between the simulated PET image generated by the next virtual scanning and the real clinical PET image.

[0075] Step six, return to step four, repeat the simulation reconstruction, and perform multiple iterations until the difference between the simulated PET image and the real PET image is less than the preset convergence threshold, or the maximum number of iterations is reached, and the optimal high-resolution activity distribution is obtained. The graph is considered to be an approximate "real" brain activity distribution graph.

[0076] The selection of the iterative optimization algorithm can be diverse, such as gradient-based optimization methods or iterative update strategies based on the expectation maximization (EM) idea, the goal of which is to minimize the difference between the simulation reconstruction result and the real clinical scanning result.

[0077] Step seven, convert the optimal "real" brain activity distribution into a "standardized PET image" by applying Gaussian filtering processing (standardization processing - that is, effective spatial resolution);

[0078] The standardization requirement usually refers to the spatial resolution of the image should reach a certain recognized standard; such as the resolution recommended by the European Nuclear Medicine Society EARL standard, or the uniform resolution defined in other multi-center research protocols.

[0079] The target of standardization (e.g. target spatial resolution FWHM value) needs to be defined in advance according to industry standards, multi-center study protocols or specific clinical requirements.

[0080] Gaussian smoothing is a commonly used and effective method to degrade the high resolution approximate "true" activity map to the image that meets the specific lower resolution standard.

[0081] Step eight, construct different candidate combinations of reconstruction parameters, and take the optimal "true" brain activity distribution as the input source image, traverse each candidate combination of reconstruction parameters combined with the system parameters of the PET device, perform simulated reconstruction to generate the corresponding second simulated PET image for each combination;

[0082] The traversal range and step size of the reconstruction parameters should be reasonably set according to the device characteristics, clinical experience and computing resources, so as to find the optimal or near-optimal parameter combination within an acceptable computing time; set a search space for the PET reconstruction parameters to be optimized, including different reconstruction algorithms (such as OSEM, FBP), different combinations of iteration times and subset arrays, different post-processing filter types and parameters (such as the FWHM of the Gaussian filter), etc. For each candidate combination of reconstruction parameters in this parameter space, use the approximate "true" brain activity distribution as the input source image, combine the current candidate reconstruction parameters and the system parameters of the PET device, and perform virtual scanning (simulated reconstruction) again to generate a new simulated PET image (referred to as the second simulated PET image).

[0083] Step nine, compare the second simulated PET image obtained by simulated reconstruction of each candidate combination of reconstruction parameters with the "standardized PET image", calculate the difference or similarity between the two, and select the candidate combination of reconstruction parameters with the smallest difference or the highest similarity as the standardized reconstruction parameters to be used for this type of clinical brain scan by the PET device.

[0084] The difference minimization criterion can be the norm of the image pixel value difference (such as L2 norm, i.e. root mean square error RMSE), the image structure similarity index (such as SSIM), or the difference of quantitative indicators in a specific region of interest, etc.

[0085] The present application starts from obtaining clinical PET and paired MRI data of patients, goes through data preprocessing, uses MRI prior to perform partial volume correction, solves the approximate "true" brain activity distribution through iterative virtual scanning, and finally determines the PET reconstruction parameters that meet the standardization requirements based on this "true" activity distribution and through parameterized virtual scanning again, to solve the technical problem of how to determine the brain PET (positron emission tomography) standardized reconstruction parameters using clinical data without the need for a physical phantom.

[0086] Embodiments:

[0087] a) Clinical data acquisition, pre-processing and extraction of anatomical constraints:

[0088] Acquire clinical brain PET image data of a patient, acquire high-resolution MRI (magnetic resonance imaging) image data paired with the PET image, register the PET image to the space of the MRI image, and segment different brain regions based on the MRI image to obtain anatomical constraints;

[0089] b) Iterative solution of an approximate "true" brain activity distribution:

[0090] Use the brain region segmentation results obtained in step a) as constraints to perform partial volume correction on the registered PET image, and based on the activity distribution after the partial volume correction as the initial estimate, combine the corresponding reconstruction parameters and PET device system parameters during PET acquisition to perform a first virtual scan to generate a first simulated PET image. Adjust the "true" activity distribution through an optimization algorithm to minimize the difference between the first simulated PET image and the clinical brain PET image, thereby obtaining an approximate "true" brain activity distribution;

[0091] c) Determine the standardization reconstruction parameter step:

[0092] Based on the approximate "true" brain activity distribution obtained in step b), generate a standardized target PET image that meets the pre-defined standardization requirements, then iterate through multiple sets of candidate reconstruction parameters in a pre-set range. For each set of candidate reconstruction parameters, use the approximate "true" brain activity distribution and combine the current candidate reconstruction parameters and the PET device system parameters to perform a second virtual scan to generate a second simulated PET image. Select the candidate reconstruction parameter combination that minimizes the difference between the second simulated PET image and the standardized target PET image as the standardization reconstruction parameter of the PET device.

[0093] The method for determining brain PET standardization reconstruction parameters based on clinical data according to the present application comprises a processing module as follows:

[0094] A clinical data acquisition, pre-processing and anatomical constraint extraction module for performing the operations of step a);

[0095] An approximate "true" brain activity distribution iterative solution module for performing the operations of step b);

[0096] A standardization reconstruction parameter determination module for performing the operations of step c) of claim 1.

[0097] The specific implementation method of "virtual scanning" or "simulated reconstruction" (such as based on Monte Carlo simulation or based on analytical projection / back-projection and system response model) is not the core of the present application as long as it can effectively simulate the PET imaging process.

Claims

1. A method for determining standardized reconstruction parameters of brain PET based on clinical brain MRI and PET data, characterized in that: The specific steps are as follows: Step 1: Given the initial reconstruction parameters, collect the patient's clinical PET scan images and paired T1W-MRI images, and perform brain structure segmentation on the T1W-MRI images to obtain MRI segmentation results; Step 2: Using the partial volume correction algorithm, the MRI segmentation results are used to correct the partial volume effect of the registered clinical PET scan images; Step 3: Map the activity values ​​from the partial volume corrected PET scans onto a high-resolution MRI spatial grid based on the brain regions defined by MRI segmentation, providing an initial estimate of the "true" brain activity distribution. Step 4: Based on the current "real" brain activity distribution map, combined with the initial reconstruction parameters and the system parameters of the PET device, a PET image simulation reconstruction process is performed to generate a first simulated PET image; Step 5: Compare the first simulated PET image with the actual clinical PET scan image, calculate the difference or similarity between the two, and use an optimization algorithm to adjust the activity values ​​in the "real" brain activity distribution map based on the difference; Step 6: Return to step 4 and repeat the simulation reconstruction for multiple iterations until the difference between the simulated PET image and the real PET image is less than a preset convergence threshold, or the preset maximum number of iterations is reached, thereby obtaining the optimal "real" brain activity distribution map; Step 7: Convert the optimal "true" brain activity distribution map into a "standardized PET image" by applying Gaussian filtering; Standardization requirements usually mean that the spatial resolution of images should meet recognized standards; Step 8: Construct different candidate combinations of reconstruction parameters, use the optimal "real" brain activity distribution map as the input source image, traverse each set of candidate reconstruction parameters and combine them with the system parameters of the PET device to perform simulation reconstruction, and generate the corresponding second simulated PET image for each set; Step 9: Compare each set of corresponding second simulated PET images with the "standardized PET images," calculate the difference or similarity between the two, and select the set of candidate reconstruction parameters with the smallest difference or highest similarity. This set of candidate reconstruction parameters is the standardized reconstruction parameters that should be used by this PET device for this type of clinical brain scan.

2. The method for determining standardized reconstruction parameters of brain PET based on clinical brain MRI and PET data according to claim 1, wherein: The reconstruction parameters include iterative algorithm type, number of iterations, number of subsets, filter type and cutoff frequency; Brain structure segmentation refers to accurately segmenting the main brain tissue areas of gray matter GM, white matter WM, cerebrospinal fluid CSF, or brain areas defined by more detailed brain maps.

3. A method for determining standardized reconstruction parameters of brain PET based on clinical brain MRI and PET data, characterized in that: The system parameters of the PET device include system geometry, detector response function / point spread function model, attenuation map and scatter correction model.

4. A method for determining standardized reconstruction parameters of brain PET based on clinical brain MRI and PET data, characterized in that: The algorithm for calculating the difference or similarity includes: mean square error, normalized mutual information or structural similarity index as the objective function.

5. A method for determining standardized reconstruction parameters of brain PET based on clinical brain MRI and PET data, characterized in that: The goal of the adjustment is to minimize the difference between the simulated PET image generated by the next virtual scan and the real clinical PET image.

Images