Cerebral Cortex Surface Segmentation Using Graph Topology Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for segmenting the cerebral cortex surface in MRI images face challenges such as high time costs, inaccurate segmentation due to individual differences, and introduction of noise during spherical mapping, especially when dealing with damaged areas.
Innovation Solution
A method utilizing a graph network for direct segmentation in the original brain space, involving graph structure modeling, deep neural network reconstruction, and graph convolution to accurately segment the cerebral cortex surface.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spherical mapping space based method is used for segmentation, then segmentation accuracy is improved theoretically, but processing time is significantly increased and noise is introduced during mapping
Solution Approach 1:
Instead of mapping the cerebral cortex surface to a spherical space for segmentation and then mapping back, the patent inverts the approach by performing segmentation directly in the original cortical surface space using graph convolutional networks. This eliminates the time-consuming forward and backward mapping processes while maintaining segmentation accuracy through graph-based modeling of the cortical geometry.
Solution Approach 2:
The patent extracts and removes the spherical mapping step from the segmentation pipeline entirely. By taking out this intermediate transformation process, the method eliminates the associated time costs and noise introduction while preserving the essential segmentation functionality through direct graph-based operations on the cortical surface.
2Measurement precision
If spherical mapping space based method is used for segmentation, then segmentation accuracy is improved theoretically, but the method fails to handle damaged areas effectively
Solution Approach 1:
The patent inverts the processing sequence by performing segmentation in the original cortical space before any spherical mapping that could distort damaged regions. The graph convolutional network operates directly on the cortical surface geometry, allowing it to naturally handle damaged areas without the artifacts introduced by spherical projection.
Solution Approach 2:
The graph convolutional network applies local adaptive processing to different regions of the cortical surface, allowing damaged areas to be segmented with appropriate local characteristics. The graph structure preserves local geometric relationships, enabling the model to adapt to varying data quality across different brain regions.
3Ease of operation
If standard map registration based method is used, then processing is simplified, but segmentation performance is reduced due to large individual differences
Solution Approach 1:
The graph convolutional network incorporates local geometric features and individual-specific cortical geometry through the graph structure, allowing the model to adapt to individual differences while maintaining processing efficiency. Each node in the graph represents a local cortical region with its unique geometric properties.
Solution Approach 2:
The method changes the parameter representation from fixed spherical coordinates to adaptive graph-based features that capture individual cortical geometry. The graph structure and convolutional operations adapt to each subject's unique brain morphology, improving segmentation performance across individuals.
4Ease of operation
If voxel based learning method is used for segmentation, then processing is simplified, but segmentation boundary accuracy is limited due to partial volume effect
Solution Approach 1:
The patent transitions from volumetric voxel-based processing to surface-based graph processing, changing the dimensional representation from 3D voxels to 2D cortical surface meshes. This dimensional change eliminates the partial volume effect by operating directly on the cortical surface where boundaries are well-defined.
Solution Approach 2:
The method segments the cerebral cortex into discrete graph nodes representing local surface regions, with boundaries defined by the mesh structure. This segmentation approach provides precise boundary definition without the ambiguity inherent in voxel-based methods.
Data Source
AI summary
A method and a system for fully automatically segmenting a cerebral cortex surface based on a graph network, the method including: registering a cerebral magnetic resonance image to a standard template space; performing cerebral cortex surface reconstruction on the registered cerebral magnetic resonance image based on a deep neural network; calculating adjacency matrices among the vertexes of the reconstructed cerebral cortex; acquiring corresponding distinguishing features of each grid vertex of the cerebral cortex surface as corresponding feature vectors of the point so as to obtain a cerebral cortex surface segmentation result; and mapping the cerebral cortex surface segmentation result back to the original coordinate space from the standard template space. Graph structure modeling is performed on the reconstructed cerebral cortex surface, and the global topological structure features are learned based on the graph network, thus realizing the accurate segmentation of the cerebral cortex surface.