基于主动梯度引导与幅值补全的人脑基底节影像分割方法
By constructing a segmentation network with active gradient guidance and amplitude completion, the problem of inaccurate boundary contours in human brain basal ganglia image segmentation was solved. By utilizing information from amplitude and magnetization maps, the segmentation effect was improved, especially the segmentation accuracy in the presence of artifacts.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIAMEN UNIV
- Filing Date
- 2023-12-13
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, the segmentation methods for human brain basal ganglia images have difficulty improving the accuracy of the boundary contours of the segmented nuclei, especially when magnetic susceptibility artifacts are present, resulting in poor segmentation performance.
A segmentation network based on active gradient guidance and magnitude completion is adopted. By constructing a magnitude information completion module and a gradient guidance branch, and training the network in combination with a loss function, the ability to extract boundary features is improved by utilizing the information of magnitude map and magnetization map.
The magnetic susceptibility map effectively reduces the influence of artifacts, improves the segmentation accuracy of the basal ganglia region boundary contour, and enhances the model's correlation and segmentation effect in the region of interest.
Smart Images
Figure CN118096789B_ABST