Multistructure large deformation diffeomorphic brain registration.

Abstract

Whole brain MRI registration has many useful applications in group analysis and morphometry, yet accurate registration across different neuropathological groups remains challenging. Structure-specific information, or anatomical guidance, can be used to initialize and constrain registration to improve accuracy and robustness. We describe here a multistructure diffeomorphic registration approach that uses concurrent subcortical and cortical shape matching to guide the overall registration. Validation experiments carried out on openly available datasets demonstrate comparable or improved alignment of subcortical and cortical brain structures over leading brain registration algorithms. We also demonstrate that a group-wise average atlas built with multistructure registration accounts for greater intersubject variability and provides more sensitive tensor-based morphometry measurements.