Brain Image Analysis

Research Summary

Our work in brain image analysis is focused on processing of structural and diffusion brain images to (i) extract and label volume and surface representations of neuroanatomic structures, with a particular emphasis on cortical anatomy, and (ii) coregister diffusion to structural images and develop methods to allow structural connectivity modeling in the context of labelled brain anatomy. The BIG lab is working with David Shattuck and colleagues at the Ahmanson-Lovelace Brain Mapping Center at UCLA to develop BrainSuite: software that will automatically extract a representation of the inner and outer cortical surface from T1-weighted images. Subsequent analysis allows coregistration of these surfaces to an atlas either automatically (using surface curvature) or based on user-drawn sulcal landmarks. Through mapping of the cerebrum to the unit ball we can then perform a volumetric coregistration in which there is a one-to-one mapping between cortical surfaces. BrainSuite also includes the ability to correct susceptibility induced distortion in diffusion images and to coregister diffusion to structural images. The software allows diffusion tensor and ODF computation and visualization, fiber tracking, and interactive visualization of connectivity using user-defined or atlas-based ROIs. Other aspects of our on-going work include methods for intersubject coregistration of ODFs, addition of a pipeline for analysis of resting state fMRI, and extensions of our image analysis tools to accommodate lesioned brains.


Research Support