Coupled nonparametric shape and moment-based intershape pose priors for multiple basal ganglia structure segmentation

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Anatomy

Background:

• Accurate segmentation of subcortical brain structures is crucial for neurological studies.
• Neighboring anatomical structures exhibit dependencies that can be exploited for improved segmentation.
• Existing methods may not fully utilize the rich spatial and shape information of brain anatomy.

Purpose of the Study:

• To develop a new active contour-based statistical method for simultaneous volumetric segmentation of multiple subcortical brain structures.
• To incorporate prior knowledge of anatomical shapes and their interrelationships into the segmentation process.
• To enhance segmentation accuracy by modeling structural co-dependencies.

Main Methods:

• Formulated segmentation as a maximum a posteriori estimation problem.
• Incorporated statistical prior models of structure shapes and relative poses using nonparametric multivariate kernel density estimation.
• Developed an active contour-based iterative algorithm within a variational framework.
• Tested on volumetric segmentation of basal ganglia structures in magnetic resonance images.

Main Results:

• Demonstrated the method's capability for simultaneous volumetric segmentation of multiple subcortical structures.
• Presented quantitative performance analysis from 2-D and 3-D experiments.
• Showcased improved segmentation accuracy compared to several existing methods.

Conclusions:

• The proposed active contour-based statistical method effectively segments multiple subcortical brain structures.
• Leveraging shape priors and inter-structure relationships significantly enhances segmentation accuracy.
• This approach offers a principled way to integrate high-level anatomical information into image segmentation.