WARPING BRAIN MR IMAGES USING DEFORMABLE CONTOURS

Abstract

Image warping is necessary for automatic identification and labeling of different brain region. It is also needed for quantitative analysis of shape variations of brain structure. In this paper, we introduce a new warping technique for brain MRI applications. We first use a deformable contour model to extract and warp the boundaries of two brain images. A mesh grid is constructed thereafter by using a distance transformation applied to the brain boundary. Finally, each brain image is matched to the other image by mapping the mesh coordinates. We have implemented and applied the method. It generates fast and accurate results for brain MRI.