Fast And Automatic Curvilinear Reformatting Of Mr Images Of The Brain For Diagnosis Of Dysplastic Lesions

Abstract

Curvilinear reformatting is known as the best non-invasive technique for diagnosis of dysplastic lesions of the brain. It consists of computing surfaces that follow the brain's curvature at various depths, making the diagnosis possible by visual inspection of the voxel intensities on these surfaces. Traditional approaches require user intervention and present curvature artifacts. We present a new method for curvilinear reformatting that solves both problems. It uses a graph-based approach to segment the brain, extract its envelope, and compute the isosurfaces at all possible depths by euclidean distance transform. It requires no user input, no ad-hoc parameters, and takes less than 1 minute to run on a common PC. © 2006 IEEE.