Automatic human brain MRI volumetric analysis technique using EM-algorithm

Abstract

The paper presents automated volumetric analysis of human brain MR images for many applications based on the Expectation-maximization (EM) algorithm. It involves voxel labeling, counting, and calculating tissues volume. The voxel labeling requires the brain magnetic resonance image segmentation which is most commonly performed based on voxels intensity signals. A widely used method for segmentation is by creating a Gaussian Mixture Model (GMM) through the EM algorithm and the same can be used to find the tissues, class label and volumes. The experimental results are provided for volumetric analysis of automated segmentation of male and female subjects as well as normal volumes of tissue classes for verifying correctness of automated volumetric analysis and statistical inference for diagnostic applications.