Automated Brain Mapping

SAVOR

Overview

Constructing a one to one correspondence between whole brain MR image scans is a problem of critical importance in neuroimaging analyses. We have developed a framework to combine the strength of both surface-based and volumetric-based analyses for consistent, bijective data transfer between brain coordinate systems.

Methods

The figure below outlines the steps in the SAVOR pipeline:

1. Two pial surfaces
2. Upper surface registered to lower surface via image registration, with identity coordinate functions overlaid.
3. Surfaces mapped to the spherical domain via FreeSurfer mapping with coordinate functions overlaid.
4. Surfaces on spherical domain with function on lower surface mapped to that on upper surface, via Spherical Demons.
5. Cortical thickness values for upper and lower brains displayed on the domain of the lower brain.

Results

• Average surface-registration error map, quantifying the variability in label overlaps after registration:



• Mean longitudinal change in cortical thickness for an 8 subject cohort, mapped on to a common template:

Relevant publications

Gibson, A., Khan, A., Beg, M. F. 2009
A combined Surface And VOlumetric Registration (SAVOR) framework to study cortical biomarkers and volumetric imaging data
MICCAI 2009. LNCS 5761: 713-724. Poster (PDF)