Multimodal surface matching with higher-order smoothness constraints.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain Mapping

Background:

• Accurate alignment of cortical surfaces is crucial for statistical sensitivity and spatial localization in group brain imaging studies.
• Cortical surface-based alignment is generally superior to volume-based methods but challenged by individual variations in cortical folding and functional area locations.
• The Multimodal Surface Matching (MSM) tool offers flexible spherical registration for accurate surface alignment using various features.

Purpose of the Study:

• To address limitations in the original MSM regularization function that allowed excessive surface distortions.
• To propose and implement a novel regularization penalty based on strain energy to improve alignment robustness.
• To introduce constraints for smooth deformation of cortical anatomies, mitigating projection distortions inherent in spherical mapping.

Main Methods:

• Developed a new regularization penalty derived from physically relevant equations of strain energy.
• Implemented constraints to enforce smooth deformation of cortical anatomies during spherical warping.
• Tested the approach on longitudinal modeling of cortical development in neonates (31-43 weeks post-menstrual age).

Main Results:

• The new regularization penalty leads to improved and more robust alignment of multimodal imaging data.
• The proposed method increases the biological interpretability of distortion fields.
• Enhanced statistical significance in population-based analysis compared to other spherical methods was demonstrated.

Conclusions:

• The novel strain energy-based regularization and smooth deformation constraints significantly improve multimodal brain surface alignment.
• This refined MSM approach offers greater biological interpretability and enhanced statistical power for neuroimaging research, particularly in developmental studies.