Lecture notes in computer science vol:1679 pages:348-357
Medical image computing and computer-assisted intervention - MICCAI'99 location:Cambridge, UK date:September 19-22, 1999
We present a completely automated procedure for measuring left and right hemispheric asymmetry in cerebral grey and white matter volumes from NIR images using a chain of state-of-the-art image analysis algorithms. After bias correction and tissue classification, left and right hemispheres are separated by non-rigid registration to a template image in which both hemispheres have been carefully segmented. Volume renderings of each hemisphere separately demonstrate the high quality of the resulting segmentations. Because all steps in the procedure are completely automated and do not require user specified parameters, the results are highly reproducible and consistent. We present quantitative results obtained from a database of NIR images of 40 schizophrenic patients and 31 normal controls.