Multistructure large deformation diffeomorphic brain registration.

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Computational Anatomy

Background:

• Accurate whole-brain MRI registration is crucial for group analysis and morphometry.
• Challenges persist in achieving robust registration across diverse neuropathological groups.
• Structure-specific anatomical information can enhance registration accuracy and stability.

Purpose of the Study:

• To introduce and validate a novel multistructure diffeomorphic registration approach for brain MRI.
• To leverage concurrent subcortical and cortical shape matching for improved registration guidance.
• To assess the performance of this method against existing leading brain registration algorithms.

Main Methods:

• Developed a multistructure diffeomorphic registration algorithm incorporating concurrent subcortical and cortical shape matching.
• Utilized openly available neuroimaging datasets for validation experiments.
• Compared the proposed method's alignment accuracy with state-of-the-art registration techniques.

Main Results:

• Demonstrated comparable or superior alignment of subcortical and cortical brain structures compared to leading algorithms.
• Showcased improved accuracy and robustness in registration across different brain regions.
• Validated the approach on publicly accessible neuroimaging data.

Conclusions:

• The multistructure registration approach effectively improves whole-brain MRI alignment, particularly in challenging neuropathological cases.
• Group-wise atlases generated using this method better capture intersubject variability and enhance morphometry analysis sensitivity.
• This technique offers a promising advancement for neuroimaging research and clinical applications.