CN-122023429-A - Image identification method for postoperative recurrence and radiation injury of glioma

CN122023429ACN 122023429 ACN122023429 ACN 122023429ACN-122023429-A

Abstract

The invention discloses a brain glioma postoperative recurrence and radiation injury image identification method, which belongs to the technical field of medical image analysis and comprises the following steps of normalizing multi-mode image data; reconstruction of micro-vascular metabolic characteristics, radiation damage structure analysis and combined constraint image identification. Constructing local perfusion response characteristics, diffusion limited characteristics and reinforced coupling characteristics through cross-modal voxel mapping and neighborhood statistical analysis to form microvascular metabolism characterization data; meanwhile, radiation damage structure characterization data are constructed through the modes of hierarchical structure region division, necrosis cavity distribution modeling, boundary continuity analysis, surrounding tissue disturbance modeling and the like, and the invention can simultaneously describe the microvascular metabolism state and the structure damage state of a focus region, and improves the accuracy and the stability of the identification of recurrence and radiation damage images after brain glioma operation.

Inventors

REN AIJUN
Jiang Fuzhuang
WANG QINGJUN

Assignees

中国人民解放军总医院第六医学中心

Dates

Publication Date: 20260512
Application Date: 20260415

Claims (9)

1. The method for identifying the postoperative recurrence and radiation injury images of the glioma is characterized by comprising the following steps of: step S1, normalizing the multi-mode image data to obtain normalized multi-mode image data; Step S2, rebuilding micro-vessel metabolism characteristics, carrying out cross-modal voxel mapping and characteristic association analysis on the interior of a focus area and the focus edge area by adopting a cross-modal micro-vessel metabolism rebuilding improvement method based on the standardized multi-modal image data, carrying out joint characteristic modeling on functional information in different modal images by constructing local perfusion response distribution, diffusion abnormal distribution and reinforced response change relation, and forming micro-vessel perfusion activity degree characteristics, diffusion limited degree characteristics and metabolism heterogeneity distribution characteristics according to local neighborhood statistical results to obtain micro-vessel metabolism characterization data; S3, analyzing a radiation damage structure, carrying out joint analysis on the internal structural form of a focus area, the continuity of focus boundaries and the disturbance condition of peripheral brain tissue structures based on the standardized multi-mode image data, adopting a hierarchical damage structure topology analysis improvement method, constructing damage structure feature expression reflecting the damage degree of the tissue structure and the structure hierarchical relationship by identifying the necrotic cavity distribution form, the edge contour completeness degree and the edema diffusion mode, and carrying out structural damage description on the focus area according to the structural abnormality degree to obtain radiation damage structure characterization data; And S4, identifying the combined constraint image, mapping the micro-vessel metabolism characterization data and the radiation damage structure characterization data to a unified identification space, and carrying out association analysis on the functional activity state and the structure damage state of the focus area to obtain combined identification result data.
2. The method for identifying postoperative recurrence and radiation injury of glioma according to claim 1, wherein in step S1, the multimode image data are standardized, multimode medical image data in a postoperative follow-up stage are obtained, spatial registration processing, voxel scale unification processing, gray level intensity normalization processing and intracranial structure region definition processing are performed on the multimode medical image data, and focus candidate regions are positioned and cut according to an anatomical structure of an operation region, so that standardized multimode image data are obtained.
3. The method for identifying the postoperative recurrence and radiation injury image of the glioma is characterized by comprising the following steps of constructing a focus area cross-modal voxel corresponding, modeling local microvascular perfusion response, modeling diffusion limited state, analyzing characteristic deviation, reconstructing regional metabolism heterogeneity inside and outside a focus and generating microvascular metabolism characterization data; the focus area cross-modal voxel corresponding construction is carried out, signal responses of each voxel position in focus candidate areas in different modal images are mapped according to the standardized multi-modal image data, focuses are divided into focus inner areas and focus edge areas according to focus area space structures, and therefore voxel-level corresponding relations among different modal images are formed, and cross-modal voxel mapping data are obtained; The local microvascular perfusion response modeling is carried out, based on the cross-modal voxel mapping data, the perfusion modal response value of each voxel neighborhood in the focus area is subjected to statistical analysis, and a local perfusion response index is constructed by calculating the local perfusion intensity and the perfusion fluctuation degree, so that local perfusion response characteristic data representing the perfusion activity degree of the microvascular in the focus area is obtained; And modeling the diffusion limited state, carrying out statistical analysis on diffusion modal response values of each voxel neighborhood in the focus area based on the cross-modal voxel mapping data, and constructing a diffusion limited index by calculating local diffusion intensity and diffusion change degree to obtain diffusion limited characteristic data representing the diffusion limited degree of focus tissues.
4. The method for identifying postoperative recurrence and radiation injury of glioma according to claim 3, wherein in step S2, the characteristic deviation analysis is performed, a cross-modal functional characteristic model is constructed based on the enhanced modal response value, the local perfusion response characteristic data and the diffusion limited characteristic data, and characteristic deviation indexes are calculated according to the cooperative change relation among different modal characteristics, so as to obtain enhanced characteristic data representing the joint change relation among enhanced response, perfusion and diffusion characteristics; The regional metabolism heterogeneity reconstruction is carried out on the perfusion response characteristics, the diffusion limited characteristics and the strengthening characteristics of the inner region and the edge region of the focus respectively, and metabolic heterogeneity indexes are constructed according to the difference degree between the two types of regions, so as to obtain microvascular metabolism heterogeneity index data reflecting the difference of metabolic states of different regions in the focus; The microvascular metabolism characterization data are generated, and the local perfusion response characteristic data, the diffusion limited characteristic data, the strengthening characteristic data and the microvascular metabolism heterogeneity index data are summarized and constructed to obtain microvascular metabolism characterization data used for characterizing the perfusion state, the diffusion limited state and the metabolic activity state of the microvascular in a focus area.
5. The method for identifying the postoperative recurrence and radiation injury image of the glioma, according to claim 4, is characterized in that in the step S2, the microvascular metabolism characterization data specifically comprise local perfusion response characteristic data, diffusion limited characteristic data, reinforcement characteristic data and microvascular metabolism heterogeneity index data, wherein the local perfusion response characteristic data are used for characterizing blood flow perfusion levels and spatial distribution differences of the blood flow perfusion levels in focal areas, the diffusion limited characteristic data are used for characterizing diffusion abnormal states caused by change of cell densities of focal tissues, the reinforcement characteristic data are used for characterizing joint change relations among reinforcement response, perfusion and diffusion characteristics, and the microvascular metabolism heterogeneity index data are used for characterizing the difference degrees of blood flow supply and metabolic activity of different areas in the focal areas.
6. The method for identifying postoperative recurrence and radiation injury of glioma according to claim 5, wherein in step S3, the hierarchical injury structure topology analysis improvement method comprises the steps of structure hierarchical region division, necrotic cavity distribution modeling, boundary continuity analysis, surrounding tissue disturbance modeling, structure topology propagation modeling and radiation injury structure characterization data generation; The structure level region division is used for carrying out space morphological analysis on the focus candidate region based on the standardized multi-mode image data, dividing the focus into a focus inner region, a focus edge region and a surrounding tissue region according to the space structure relationship of the focus region, and constructing a focus structure level region mapping relationship to obtain structure level region mapping data; The necrotic cavity distribution modeling is carried out, based on the structural hierarchy region mapping data, the image signal intensity of voxels in the focus inner region and the neighborhood change condition thereof are analyzed, and the spatial position and the morphological distribution of the necrotic cavity are modeled and described by identifying the local signal abnormal reduction region and the communication structure thereof, so that the necrotic cavity distribution characteristic data is obtained; And the boundary continuity analysis is used for carrying out statistical analysis on the image gradient change, the edge form continuity and the local structure change condition of the focus edge area based on the structure level area mapping data, and carrying out quantitative expression on the integrity of the focus boundary outline by constructing a boundary continuity index to obtain boundary continuity characteristic data.
7. The method for identifying postoperative recurrence and radiation injury image of glioma according to claim 6, wherein in step S3, the peripheral tissue disturbance modeling is performed, based on the structural hierarchy region mapping data, statistical analysis is performed on image signal change conditions and spatial distribution characteristics in the peripheral tissue region, and modeling description is performed on peripheral brain tissue structure damage conditions by identifying abnormal signal diffusion regions and disturbance degrees thereof, so as to obtain peripheral tissue disturbance characteristic data; the structural topology propagation modeling is carried out, a topological association model between focus structure levels is constructed based on the necrotic cavity distribution feature data, the boundary continuity feature data and the surrounding tissue disturbed feature data, and structural damage propagation intensity is modeled and expressed by analyzing the spatial transfer relation of structural anomalies between different level areas, so that structural level association feature data is obtained; And generating the radiation damage structure characterization data, and summarizing and constructing the necrotic cavity distribution characteristic data, the boundary continuity characteristic data, the peripheral tissue disturbed characteristic data and the structure level associated characteristic data to obtain the radiation damage structure characterization data for characterizing the damage degree of the focus area structure and the structure level relation.
8. The method for identifying postoperative recurrence and radiation injury image of brain glioma according to claim 7, wherein the radiation injury structure characterization data comprises necrosis cavity distribution characteristic data, boundary continuity characteristic data, peripheral tissue disturbed characteristic data and structure level associated characteristic data; The necrosis cavity distribution characteristic data is used for representing the space distribution state of a low-activity necrosis area inside a focus, the boundary continuity characteristic data is used for representing the integrity degree and the local interruption condition of the boundary outline of the focus, the peripheral tissue disturbed characteristic data is used for representing the edema infiltration state and the structural damage degree of peripheral brain tissues of the focus, and the structural hierarchy association characteristic data is used for representing the morphological transfer relationship and the topological association relationship among the focus inner area, the edge area and the peripheral tissue area.
9. The method for identifying postoperative recurrence and radiation injury image of glioma according to claim 8, wherein in step S4, the combined constraint image is identified, the microvascular metabolic characterization data and the radiation injury structure characterization data are mapped to a unified identification space, the functional activity state and the structure injury state of a focus area are subjected to association analysis, candidate identification results are subjected to combined screening by constructing functional activity consistency constraint and structure injury consistency constraint, identification results which do not meet cross-mechanism consistency conditions are subjected to inhibition treatment, identification results of the focus belonging to postoperative recurrence or radiation injury of glioma are output, and combined identification result data are obtained.

Description

Image identification method for postoperative recurrence and radiation injury of glioma Technical Field The invention relates to the technical field of medical image analysis, in particular to a method for identifying postoperative recurrence and radiation injury images of glioma. Background The method for identifying the postoperative recurrence and radiation injury image of the glioma is a medical image analysis method which is used for judging pathological properties of abnormal image manifestations of an operation area by systematically analyzing multi-mode medical image data obtained in a postoperative follow-up stage of a glioma patient so as to distinguish whether a focus belongs to tissue injury caused by tumor recurrence or radiation therapy. The method has the main effects of providing an auxiliary judgment basis for a clinician by comprehensively analyzing information such as image signal change, tissue structure change, functional metabolism change and the like of a focus area, thereby improving the accuracy of postoperative follow-up evaluation. In practical application, by accurately distinguishing tumor recurrence from radiation injury, doctors can be helped to formulate more reasonable follow-up treatment strategies, such as adjusting radiotherapy schemes, developing re-operation treatment or carrying out targeted drug treatment, so that the treatment effect of patients is improved and unnecessary medical intervention is reduced. The image analysis method commonly used in clinic at present mostly depends on single-mode image characteristics or simple statistical indexes to judge, for example, auxiliary analysis is performed by enhancing MRI signal change, diffusion weighted signals or perfusion indexes. However, because different physiological mechanism information such as blood perfusion, tissue diffusion, and structural morphology changes are reflected by different modality images, if a unified feature integration mechanism is lacking, it is often difficult to comprehensively describe the true pathological state of the focal region. Meanwhile, the existing method is usually insufficient in attention to lesion structural damage characteristics, and is difficult to systematically describe hierarchical relationships among necrotic structures, boundary damages and peripheral tissue disturbances, so that misjudgment is easy to generate in complex cases; Therefore, how to synthesize the microvascular metabolic activity characteristics and the structural damage characteristics of the focus area on the basis of the multi-mode medical image data and construct a stable and reliable combined discrimination mechanism, so as to realize the accurate discrimination of the postoperative recurrence and the radiation damage of the glioma, and the method becomes a technical problem to be solved in the field of the auxiliary diagnosis of the current medical image. Disclosure of Invention Aiming at the situation, the invention provides a method for identifying images of recurrence and radiation injury after brain glioma operation, which aims to overcome the defects of the prior art, and the technical scheme adopted by the invention is as follows: step S1, normalizing multi-mode image data; s2, reconstructing the metabolic characteristics of the micro blood vessels; s3, analyzing a radiation damage structure; And S4, joint constraint image authentication. Further, in step S1, the multi-mode image data is normalized, and is used for performing normalization processing on a multi-mode medical image of a patient after a glioma operation, specifically, obtaining multi-mode medical image data in a post-operation follow-up stage, performing spatial registration processing, voxel scale unification processing, gray level intensity normalization processing and intracranial structure region definition processing on the multi-mode medical image data, and performing positioning and clipping processing on a focus candidate region according to an anatomical structure of an operation region to obtain normalized multi-mode image data. Further, in step S2, the microvascular metabolic feature reconstruction is configured to reconstruct, based on a standardized image expression basis, a composite image feature reflecting a microvascular perfusion state, a tissue diffusion state and a local metabolic activity state of a focal region, specifically, based on the standardized multi-modal image data, perform cross-modal voxel mapping and feature association analysis on the interior of the focal region and a focal edge region by using a cross-modal microvascular metabolic reconstruction improvement method, perform joint feature modeling on functional information in different modal images by constructing a local perfusion response distribution, a diffusion abnormal distribution and an enhanced response variation relationship, and form microvascular perfusion activity degree feature, diffusion limited degree feature and metabolic heterog