Rapid automatic brain volumetry on the basis of multispectral 3D MR imaging data on personal computers

Abstract

Fast magnetic resonance imaging sequences are an excellent basis for the volumetry of the human brain. We present a concept for an automatic three-dimensional (3D) segmentation and volumetry which provides rapid image processing along with minimal requirements of manual interaction on low-end hardware platforms. The concept is based on standard image segmentation techniques and provides for the determination of the total brain tissue and cerebrospinal fluid volumes from the input data base, a T1- and a T2-weighted 3D image data cube. We used a 3D MPRAGE sequence for the T1- and a 3D PSIF sequence for the T2-weighted images; both 3D image data cubes consisted of 128 sagittal image slices with a voxel size of 1.0 x 1.0 x 1.2 mm3. The segmentation and volumetry tool was applied to clinical image data of volunteers and Alzheimer patients in order to prove its stability.