Geodesic closest point constrained inter-subject non-rigid registration

Abstract

In this paper, we propose an inter-subject brain registration method by combining the intensity and the geodesic closest point based similarity metric. Each of the brain hemispheres can be topologically equalized into a sphere. The inter subject variance can be better reduced by using cortical surface based analysis method. A one to one mapping of the points on spherical surfaces of two subjects can be achieved by using this technique. We find the geodesic correspondence between subjects by using spherical registration first, then the correspondence on the cortical surface between subjects are used to guide the volumetric inter-subject registration. By adding these anatomical constraints of the cortical surface, the inter-subject registration result is more anatomically meaningful and accurate. The cortical surface correspondence between subjects can be combined with the general non-rigid registration. In our experiments, the proposed method performs better than the method of Hartkens et al.