Deformation Morphometry and Registration
The Deformation Morphometry and Registration project unifies information theory and continuum mechanics to track brain changes in MRI and DTI scans.
Deformation morphometry is arguably the most powerful approach ever developed to detect and analyze longitudinal changes in the living brain. This project takes a complementary approach to the Image Processing and Segmentation project’s cortical modeling approaches by developing nonlinear deformation programs with extraordinary power to track subtle volumetric changes throughout the brain. Elastic and fluid registration deformation techniques can automatically track brain changes over time in thousands of MRI and DTI brain scans, revealing the complex 3D profiles of tissue gain or loss, population differences, and time-dependent changes. Deformation maps, with millions of degrees of freedom, are used to detect and analyze longitudinal changes in the living brain. These approaches reveal how diseases develop and spread in the living brain, what factors influence their expression, and how treatments resist them.
The Deformation Morphometry and Registration Project has the following aims:
- To validate and optimize fluid registration using new information-theoretic measures for 3D fluid registration of brain images that show promise relative to the gold standard (mutual information-based registration).
- To improve the mathematics of tensor-based morphometry (using Lie group statistics to improve power, and large-deformation Riemannian elasticity and information distances to remove bias) by testing three mathematical innovations that analyze the full tensor-valued information in deformation morphometry.
- To perform information-theoretic fluid registration of DTI (diffusion tensor images) by testing a fluid image registration approach to align and analyze DTI data, with innovations to accommodate High Angular Resolution Diffusion Images (HARDI).
- To develop a processing sequence to map brain changes in longitudinal MRI (N=1000+ scans) with tensor-based morphometry (TBM).
Taken together, these tools allow us to detect morphometric changes in longitudinal MRI and DTI data, with sufficient automation to process thousands of images with unprecedented speed and power.