SurfNet: Reconstruction of Cortical Surfaces via Coupled Diffeomorphic Deformations

Area of Science:

• Neuroimaging
• Computational Anatomy
• Medical Image Analysis

Background:

• Accurate reconstruction of cortical surfaces from brain MRIs is crucial for understanding brain structure and function.
• Existing methods often reconstruct surfaces independently, neglecting their interdependence and potentially leading to inaccuracies.
• Estimating cortical thickness relies heavily on precise surface reconstruction.

Purpose of the Study:

• To develop a novel method for fast and accurate joint reconstruction of inner, outer, and midthickness cortical surfaces from brain MRIs.
• To leverage the interdependence of these surfaces for improved reconstruction accuracy and topological correctness.
• To enable precise cortical thickness estimation through coupled surface deformation analysis.

Main Methods:

• Jointly reconstructs inner (white-gray matter interface), outer (pial), and midthickness surfaces using three coupled diffeomorphic deformations.
• Optimizes midthickness surface to lie halfway between inner and outer surfaces.
• Employs regularization terms for non-negative cortical thickness and symmetric cycle-consistency to ensure spherical topology.

Main Results:

• Achieves state-of-the-art performance in cortical surface reconstruction accuracy.
• Demonstrates superior surface topological correctness compared to existing methods.
• Validated on large-scale MRI datasets including ADNI, HCP, and OASIS.

Conclusions:

• The proposed joint reconstruction method offers significant improvements in accuracy and topological correctness for brain surface analysis.
• This approach facilitates more reliable estimation of cortical thickness.
• The method shows promise for large-scale neuroimaging studies and clinical applications.