Statistical shape analysis of subcortical structures using spectral matching.

Area of Science:

• Neuroimaging
• Computational Anatomy
• Medical Image Analysis

Background:

• Morphological changes in subcortical structures are key indicators of neurodevelopmental and neurodegenerative diseases.
• Accurate quantification of these changes is crucial for understanding and diagnosing neurological disorders like Alzheimer's disease and schizophrenia.
• Existing groupwise shape analysis methods require improvement in accuracy and efficiency.

Purpose of the Study:

• To propose a novel groupwise shape analysis approach for detecting regional morphological alterations in subcortical structures.
• To enhance the accuracy of surface correspondence using mean curvature features.
• To validate the method's performance and clinical applicability in identifying pathological brain changes.

Main Methods:

• Extraction of smoothed triangulated surface meshes from segmented binary maps.
• Establishment of point-to-point correspondences using spectral matching with mean curvature features.
• Generation of mean shapes for each group and creation of a distance map to quantify morphological differences.
• Statistical analysis of the distance map to identify significant inter-group variations.

Main Results:

• The proposed method successfully detects regional morphological alterations in subcortical structures (hippocampus, putamen).
• It demonstrates comparable performance to state-of-the-art algorithms with reduced computational cost.
• The method shows high sensitivity in identifying subtle morphological variations in Alzheimer's disease patients.

Conclusions:

• The novel groupwise shape analysis framework provides an accurate and efficient tool for studying subcortical structure morphology.
• It holds significant potential for clinical applications in the early detection and diagnosis of neurological disorders.
• Incorporating mean curvature features improves the reliability of surface correspondence in shape analysis.