Large Deformation Multiresolution Diffeomorphic Metric Mapping for Multiresolution Cortical Surfaces: A...

Area of Science:

• Neuroimaging
• Computational Anatomy
• Medical Image Analysis

Background:

• Brain surface registration is crucial for studying cortical variations in development and disease.
• Current methods face challenges due to complex anatomy and inter-individual differences.

Purpose of the Study:

• To propose a fast, accurate coarse-to-fine algorithm for brain surface registration.
• To improve upon the Large Diffeomorphic Deformation Metric Mapping (LDDMM) framework for surface mapping.

Main Methods:

• Adapted LDDMM using multiresolution analysis of surface meshes.
• Constructed multiresolution diffeomorphic transformations and deformation vector fields.
• Employed a coarse-to-fine strategy for computational efficiency.

Main Results:

• Demonstrated improvements in speed and accuracy compared to standard LDDMM.
• Achieved better sulcal alignment through multiresolution mesh priors.
• Reduced computational cost via the coarse-to-fine approach.

Conclusions:

• The proposed multiresolution LDDMM algorithm offers a faster and more accurate solution for brain surface registration.
• This method advances the characterization of cortical anatomy and its variations.
• Potential applications include understanding neurodevelopment and psychiatric conditions.