MRI segmentation method based on reinforcement learning multi-scale neural network

Abstract

The invention discloses an MRI segmentation method based on a reinforcement learning multi-scale neural network, and solves the problems of error-prone segmentation and missing segmentation of multi-scale, fuzzy and diffuse MRI inflammatory regions in the existing method. According to the method, the voxel constraint strategy of modifying the inflammatory region label value according to the voxelvalue is adopted, and the segmentation effect of the segmentation model on the inflammatory region with large voxel value difference is improved; according to the method, a multi-scale convolution module GMS is designed for inflammatory regions with large shape and size differences, and the segmentation effect of the segmentation model on the multi-scale inflammatory regions is improved; in orderto solve the problem that the segmentation model is difficult to identify the diffusion blurred inflammatory region, the reinforcement learning network is used for data enhancement, and the discrimination performance of the segmentation model for the diffusion blurred inflammatory region is improved. According to the method, wrong segmentation and missing segmentation of the multi-scale fuzzy diffusion inflammatory region are reduced, and the segmentation effect of the MRI inflammatory region is improved. The method can be used for MRI inflammatory region segmentation and quantitative analysisautomatic processing.

Description

technical field [0001] The invention belongs to the technical field of image processing, and mainly relates to the single-mode image segmentation of nuclear magnetic resonance inflammatory lesions, specifically an MRI segmentation method based on reinforcement learning multi-scale neural network, which can be used for the inflammation area in hip joint nuclear magnetic resonance image MRI Data partitioning. Background technique [0002] Ankylosing spondylitis (AS) comprises a group of related disorders characterized by inflammation of the sacroiliac joints and at sites such as the spine, peripheral joints, and tendon attachments. The etiology of AS is complex, the pathogenesis is unclear, the early clinical manifestations of patients are not typical, the traditional radiological examination is not sensitive, and the lack of specific laboratory indicators has caused great obstacles for clinicians to accurately judge AS in the early stage. In time, it will often cause serious...

AI Technical Summary

Problems solved by technology

[0004] Because the inflammatory area is irregular in shape, size, and distribution on the MRI data, there are problems such as inability to quantify and inconsistency in judging the severity when clinicians use MRI data to analyze patients. The difference in the quality of MRI data caused by the difference of different equipment increases the difficulty of reading images for clinicians

[0005] Existing medical image segmentation algorithms have serious missing and mis-segmented problems when segmenting inflammatory regions, which seriously affects doctors' judgment and quantitative analysis of patients, leading to wrong treatment plans. Some scholars published in Magnetic Resonance In Medicine 2019 The Measurement of synovial tissue volume in kneeosteoarthritis using a semiautomated MRI-based quantitative approach uses the prior information of the inflammatory region data and the threshold segmentation method to realize the segmentation of the inflammatory region. However, due to the lack of consideration of the multi-scale characteristics of the inflammatory region, the inflamed region poor segmentation

Images