Groupwise registration by hierarchical anatomical correspondence detection.

Area of Science:

• Medical Imaging
• Computational Anatomy
• Neuroimaging

Background:

• Accurate alignment of multiple brain images is crucial for group studies.
• Existing groupwise registration methods face challenges with complexity and accuracy.

Purpose of the Study:

• To develop a novel feature-based groupwise registration method for simultaneous alignment of subjects.
• To improve robustness and accuracy in anatomical correspondence detection and deformation field estimation.

Main Methods:

• Decoupling groupwise registration into correspondence detection and dense deformation interpolation.
• Utilizing attribute vectors as morphological signatures for robust anatomical correspondence.
• Employing soft correspondence assignment and thin-plate splines for accurate transformation estimation.

Main Results:

• The proposed method demonstrated more robust and accurate registration results compared to existing groupwise and pairwise methods.
• Evaluated on diverse datasets including elderly brains, NIREP, and LONI data.

Conclusions:

• The novel feature-based groupwise registration method effectively aligns multiple subjects to a common space.
• The approach offers significant improvements in accuracy and robustness for neuroimaging analysis.