Automatic segmentation of 3D-MRI data using a genetic algorithm

Abstract

One of the most interesting recently developed brain activity imaging methods is functional MR imaging (fMRI). The advantages of fMRI, i.e. noninvasiveness, reproducibility and interactivity of examination, must be measured against the problems like data distortion and limited time for examination. A major problem is that most fMRI segmentation procedures are partly interactive. There is a high demand for precisely and automatically working segmentation algorithms in order to get meaningful results within an acceptable short time. This article discusses the use and implementation of a genetic algorithm (GA) as a kernel for an automatic 3D segmentation of gray matter and white matter of a human brain within the procedure of fMRI.