Search

CN-121981951-A - Method and system for detecting PET and MRI epilepsy induction zone of epileptic patient

CN121981951ACN 121981951 ACN121981951 ACN 121981951ACN-121981951-A

Abstract

The application discloses a method and a system for detecting PET and MRI epileptic areas of epileptic patients. The method comprises the steps of obtaining a preprocessed 3D T1-BRAVO sequence image, a preprocessed rs-fMRI sequence image and a preprocessed 18F-FDG-PET sequence image of a patient, generating a Z value graph group according to the preprocessed 3D T1-BRAVO sequence image, obtaining a structural abnormality region marking result corresponding to the 3D T1-BRAVO sequence according to the Z value graph group, generating a glucose metabolism abnormality region marking result corresponding to the 18F-FDG-PET sequence according to the preprocessed 18F-FDG-PET sequence image, and generating high-confidence induction focus region positioning information according to the preprocessed rs-fMRI sequence image. The detection method for the PET and MRI epileptic areas of the epileptic patients fuses three-mode data of 3D T1-BRAVO (structure), rs-fMRI (function) and 18F-FDG-PET (metabolism), breaks through the limitation of single-mode information, innovatively adopts pBFS technology to construct an individual brain function template containing 213 function areas, and thoroughly overcomes the defect that the traditional group template ignores individual differences.

Inventors

  • WANG JINGJUAN
  • LU JIE
  • YIN YAYAN
  • ZHANG YUE

Assignees

  • 首都医科大学宣武医院

Dates

Publication Date
20260505
Application Date
20251224

Claims (10)

  1. 1. The PET and MRI epilepsy-inducing region detection method for the epileptic patients is characterized by comprising the following steps of: Acquiring a preprocessed 3D T1-BRAVO sequence image, a preprocessed rs-fMRI sequence image and a preprocessed 18F-FDG-PET sequence image of a patient; Generating a Z-value graph group according to the preprocessed 3D T1-BRAVO sequence images; obtaining a structure abnormal region marking result corresponding to the 3D T1-BRAVO sequence according to the Z value graph group; Generating a glucose metabolism abnormal region marking result corresponding to the 18F-FDG-PET sequence according to the preprocessed 18F-FDG-PET sequence image; and generating high-confidence-degree epileptogenic region positioning information according to the preprocessed rs-fMRI sequence image.
  2. 2. The method for PET and MRI for a epileptic field detection of an epileptic patient according to claim 1, wherein said generating a set of Z-value maps from the preprocessed 3d T1-BRAVO sequence of images comprises: obtaining a trained U-Net++ cortex segmentation network; Inputting the preprocessed 3D T1-BRAVO sequence image into a U-Net++ cortex segmentation network, so as to obtain a GM-WM layer mask image, a GM-Surface layer mask image and a cortex column core region mask image; respectively carrying out characteristic enhancement on the GM-WM layer mask image, the GM-Surface layer mask image and the cortical column core region mask image, so as to obtain an enhanced GM-WM layer image, an enhanced GM-Surface layer image and an enhanced cortical column core region image; Performing connection point parameter calculation according to the enhanced GM-WM layer image, the enhanced GM-Surface layer image and the enhanced cortex column core region image, so as to obtain a connection original image; performing expansion parameter calculation according to the enhanced GM-WM layer image and the enhanced GM-Surface layer image, so as to obtain an Extension original image; generating a Theckness original image according to the GM-WM layer mask image, the GM-Surface layer mask image and the preprocessed 3D T1-BRAVO sequence image; acquiring parameter information of a reference area; and generating a Z-value graph group according to the reference area parameter information, the connection original graph, the Extension original graph and the Thickness original graph.
  3. 3. The method for detecting PET and MRI epileptogenic regions of an epileptic patient according to claim 2, wherein the obtaining the structure abnormality region labeling result corresponding to the 3d T1-BRAVO sequence according to the Z-value map group comprises: acquiring a Z value threshold; and marking each region with the Z value larger than the Z value threshold value in the Z value graph group as a structural abnormal region, thereby obtaining a structural abnormal region marking result.
  4. 4. The method for detecting PET and MRI seizures by using epileptic patients as claimed in claim 3, wherein said generating the labeling result of abnormal glucose metabolism region corresponding to 18F-FDG-PET sequence based on the preprocessed 18F-FDG-PET sequence image comprises: And carrying out partial volume effect correction on the PET image by adopting a QPET quantitative analysis method, calculating a standardized uptake value ratio, and comparing the data with healthy control group data through double-sample t-test to identify regions for increasing or reducing abnormal glucose metabolism, thereby obtaining a labeling result of the abnormal glucose metabolism region and a PET SUVR quantitative graph.
  5. 5. The method for PET and MRI for epileptic area detection of an epileptic patient according to claim 4, wherein said high confidence epileptic focus area localization information based on the preprocessed rs-fMRI sequence images comprises: dividing the brain of an individual into 213 functional areas based on the preprocessed rs-fMRI sequence image by pBFS technology, and generating an individual brain function template, thereby obtaining the individual brain function template with 213 functional area divisions; mapping the structure abnormal region marking result and the glucose metabolism abnormal region marking result to an individual brain function template to form an individual pathogenic involvement template; calculating a functional connection matrix of the preprocessed rs-fMRI sequence image on an individual pathogenic involvement template, thereby obtaining an individual epilepsy network; Generating an epilepsy network key node, a MAP three-group parameter image probability heat MAP, a PET parameter image probability heat MAP and a multi-mode epilepsy network key node ordering probability heat MAP according to the individual epilepsy network; And generating high-confidence-degree epileptogenic region positioning information according to the MAP three-group parameter image probability heat MAP, the PET parameter image probability heat MAP and the multi-mode epileptogenic network key node ordering probability heat MAP.
  6. 6. The method for PET and MRI for a epileptic region detection according to claim 5, wherein said dividing the individual brain into 213 functional regions based on the preprocessed rs-fMRI sequence image using pBFS technique, generating an individual brain function template, thereby obtaining an individual brain function template having 213 functional region divisions comprises: Performing anatomical guide pretreatment on the pretreated rs-fMRI sequence image, thereby obtaining a gray matter region triple mask map, a denoised rs-fMRI gray matter region time sequence and brain lobe partition grouping data; generating 213 individualized seed point coordinate sets according to brain leaf partition grouping data; generating 213 preliminarily divided functional area mask diagrams according to 213 personalized seed point coordinate sets, the denoised rs-fMRI gray matter area time sequence and the gray matter area triple mask diagram; and generating an individualized brain function template containing 213 functional area partitions according to the 213 preliminarily partitioned functional area mask graphs.
  7. 7. The method of claim 6, wherein mapping the result of the structural abnormality region labeling and the result of the glucose metabolism abnormality region labeling onto an individual brain function template to form an individual pathogenic involvement template comprises: respectively carrying out coordinate standardization processing on the structure abnormal region marking result and the glucose metabolism abnormal region marking result, thereby obtaining a structure abnormal region mask graph after coordinate calibration and a glucose metabolism abnormal region mask graph after coordinate calibration; carrying out functional area-abnormal area space overlapping quantitative judgment on the structural abnormal area mask graph after coordinate calibration and the glucose metabolism abnormal area mask graph after coordinate calibration and the individualized brain function template, thereby obtaining a functional area-abnormal involvement comparison table; generating a functional area-involvement grade mapping table and a functional area division map with grade labels according to the functional area-involvement comparison table, the structural abnormal area mask map after coordinate calibration and the glucose metabolism abnormal area mask map after coordinate calibration; Generating individual pathogenic involvement templates according to the function division map with the class labels, the individual brain function templates with 213 function division and the function region-involvement class mapping table.
  8. 8. The method of claim 7, wherein calculating a functional connection matrix of the preprocessed rs-fMRI sequence images on the individual disease-causing involvement template to obtain the personalized epileptic network comprises: Generating core signal segments of 13 functional areas and signal segment extraction parameter tables of each new number segment according to the preprocessed rs-fMRI sequence image and the individual pathogenic involvement template; Performing classification type accumulated functional area grouping and signal segment association according to the core signal segment and the individual pathogenic accumulated template, so as to generate an A group functional area grouping table, a B group functional area grouping table, a C group functional area grouping table and a core signal segment set after grouping association; generating a group A functional connection strength matrix, a group B functional connection strength matrix and a group C functional connection strength matrix according to the group A functional area grouping table, the group B functional area grouping table, the group C functional area grouping table and the core signal segment set after grouping association; generating epilepsy network skeleton data according to the A group functional connection strength matrix, the B group functional connection strength matrix and the C group functional connection strength matrix; generating individual epilepsy-inducing network according to the skeleton data of the epilepsy-inducing network and the individual disease-causing involvement template.
  9. 9. The method for PET and MRI epileptic area detection of an epileptic patient according to claim 8, wherein said generating an epileptic network key node, a MAP three-set of parameter image probability heat MAPs, a PET parameter image probability heat MAP, a multi-modal epileptic network key node ordering probability heat MAP according to the personalized epileptic network comprises: Screening key nodes of the epilepsy induction network according to the individual epilepsy induction network and the individual pathogenic involvement template, thereby obtaining a list of key nodes of the epilepsy induction network and a space marking chart of the key nodes; Generating a Junction probability heat MAP, an Extension probability heat MAP and a Thickness probability heat MAP according to the key node space mark MAP, the Junction original MAP, the Extension original MAP, the Thickness original MAP and the structure anomaly area mark result, wherein the Junction probability heat MAP, the Extension probability heat MAP and the Thickness probability heat MAP form three groups of parameter image probability heat MAPs; Generating a PET parameter image probability heat map according to the glucose metabolism abnormal region marking result, the PET SUVR quantitative map and the key node space marking map; generating a multi-mode epilepsy induction network key node ordering probability heat map according to the epilepsy induction network key node list, the connection probability heat map, the Extension probability heat map, the thermal probability heat map, the PET parameter image probability heat map and the personalized brain function template.
  10. 10. A PET and MRI epileptic area detection system for an epileptic patient, the PET and MRI epileptic area detection system for an epileptic patient comprising: the preprocessing module is used for acquiring preprocessed 3D T1-BRAVO sequence images, preprocessed rs-fMRI sequence images and preprocessed 18F-FDG-PET sequence images of a patient; the Z-value image group generation module is used for generating a Z-value image group according to the preprocessed 3D T1-BRAVO sequence images; the structure abnormal region marking result generation module is used for acquiring a structure abnormal region marking result corresponding to the 3D T1-BRAVO sequence according to the Z value graph group; The glucose metabolism abnormal region marking result generation module is used for generating a glucose metabolism abnormal region marking result corresponding to the 18F-FDG-PET sequence according to the preprocessed 18F-FDG-PET sequence image; the device comprises a high-confidence-degree epileptic focus area positioning information generation module, wherein the high-confidence-degree epileptic focus area positioning information generation module is used for generating high-confidence-degree epileptic focus area positioning information according to the preprocessed rs-fMRI sequence image.

Description

Method and system for detecting PET and MRI epilepsy induction zone of epileptic patient Technical Field The application relates to the technical field of image processing, in particular to a method and a system for detecting PET and MRI induced epilepsy areas of epileptic patients. Background Epilepsy is a recurrent chronic disease of the nervous system that severely threatens human health. WHO statistics show that the prevalence rate of epilepsy is 5-11.2 per mill, about 5000 thousands of epileptic patients exist in the whole world, about 900 thousands of epileptic patients in China, including 600 thousands of active epileptic diseases, and about 40 thousands of new cases are added each year. About 20-30% of all epileptic patients are refractory to drugs, surgical treatment mainly including surgical excision and neuromodulation is needed, but about 30% of patients relapse after surgery. Studies have shown that epilepsy is a network disease whose onset and spread depends on a specific pathogenic network. Traditional treatment concepts with 'pathogenic focus' as a core fail to fully consider the integrity of the network and key nodes, resulting in poor treatment effect for some patients. Magnetic resonance imaging (Magnetic Resonance Imaging, MRI) has the advantages of high resolution, abundant information, no ionizing radiation damage and the like, is a primary method for checking the etiology of epilepsy, and partial hippocampal sclerosis and temporal lobe lesions can be seen on a conventional MRI sequence, but about 1/3 of temporal lobe epilepsy is MRI negative, and clinical diagnosis is difficult. The image post-processing technology of the structural image can be used for measuring the structural image characteristics of the nuclear magnetic negative temporal lobe epilepsy. Based on a morphological analysis method (morphometric analysis program, MAP) of the high-resolution nuclear magnetic T1 structural image, a brain structure 'connection MAP', 'expansion MAP' and 'cortex thickness MAP' are calculated. The high definition structure T1 image analyzed by MAP can provide the information of the epileptic network node for TLE patients. PET is used as a molecular imaging technology, can provide metabolic information, and has important value for judging the range of epilepsy. Previous 18F-FDG PET studies indicate that brain glucose metabolism patterns are an important means for judging the involvement of an epileptogenic network in brain areas. Abnormal cerebral cortex discharge in epileptic seizure stage, obviously increased tissue energy consumption, local 18F-FDG hypermetabolism, opposite seizure stage, 18F-FDG hypometabolism and seizure stage hypermetabolism region, can be identified as epileptogenic focus. Applicant team has established chinese interseizure 18F-FDG PET normal think tank and has initially performed verification of locating epileptic foci. The seizure interval 18F-FDG PET visualizes the occurrence of hypometabolism in the hippocampus and temporal lobe areas, and can accurately locate the seizure causing areas of 85% -90% of temporal lobe epileptic patients. The applicant has mastered the Quantitative PET (QPET) method, measured the hypometabolic characteristics of the lesion by means of standardized uptake values, and studied its value of localization to TLE patients The rs-fMRI based on BOLD does not need to complete the appointed task, and provides the low-frequency oscillation time sequence of each abnormal brain region, so that the method is suitable for the research of constructing the temporal lobe epilepsy-induced epilepsy network. The centrality (DEGREE CENTRALITY, DC) parameter based on graph theory quantifies the importance of a node in an epileptic network. The Brain magazine builds a network for inducing epilepsy, compares the network DC value of prognosis of the excision operation, finds that the DC value in the excision area of the group with good prognosis is higher than the excision range, and confirms the role of DC in positioning the network for inducing epilepsy and evaluating prognosis. The method for positioning the epilepsy induction region based on the single-mode structure MRI, fMRI and FDG PET has the defects of single-mode information, neglects the difference of individual brain function structures of patients based on group templates, does not build an epilepsy induction network from network dimension to search key nodes, and is difficult to effectively guide individual accurate treatment. Disclosure of Invention The present invention is directed to a method for detecting a PET and MRI epileptic region of an epileptic patient, which solves at least one of the above-mentioned problems. In one aspect of the present invention, there is provided a method for detecting a PET and an MRI epileptic region of an epileptic patient, the method comprising: Acquiring a preprocessed 3D T1-BRAVO sequence image, a preprocessed rs-fMRI sequence image and a preprocessed 18F-FDG-P