Ultra-fast automated brain volumetry based on bispectral MR imaging data

In this paper we present a fast automated three-dimensional brain segmentat ion and brain volumetry method providing minimal requirements of the number of spectral MR image channels and the performance of the computer equipmen t employed. The presented method is based on standard non-iterative two-dim ensional grey level segmentation techniques in combination with pixel vecto r-oriented classificators and morphological operators and requires a bispec tral high resolution MR data base. In its current design, the method provid es for the largely partial volume-corrected determination of the cerebrospi nal fluid (CSF) and the white and grey brain matter volumes of the human br ain. For the verification of our approach, the results of applications to c linical data were compared to those of studies found in the literature. (C) 1999 Elsevier Science Ltd. All rights reserved.