Cortical shape analysis in the Laplace-Beltrami feature space

Area of Science:

• Neuroscience
• Computational Anatomy
• Medical Image Analysis

Background:

• Automated analysis of cortical morphometry requires precise descriptions of anatomical structures on the complex cerebral cortex.
• Current methods may lack robustness to variations in scale and pose, hindering widespread application.

Purpose of the Study:

• To propose a novel feature space for characterizing cortical geometry.
• To develop a robust, learning-based algorithm for automated sulci detection within this new feature space.

Main Methods:

• Utilizing eigenfunctions of the Laplace-Beltrami operator to derive an intrinsic geometric feature space.
• Developing and applying a learning-based algorithm for sulci detection in the proposed feature space.

Main Results:

• The proposed feature space is invariant to scale and pose variations.
• The feature space demonstrates anatomical meaningfulness and robustness across populations.
• Successful automated sulci detection was achieved on 10 training and 15 testing cortical surfaces.

Conclusions:

• The novel feature space effectively characterizes cortical geometry for automated analysis.
• The learning-based sulci detection algorithm shows promise for advancing cortical shape analysis.
• This method offers a robust solution for analyzing brain structure variations.