Efficient Detection of Mesial Temporal Sclerosis using Hippocampus and CSF Features in MRI Images

Abstract

Mesial temporal sclerosis (MTS) is one of the most common pathological abnormalities associated with temporal lobe epilepsy. Prompt identification of MTS can determine surgical candidacy of a medically refractory epilepsy patient thus reducing morbidity and mortality. Traditionally, MTS is detected by visual inspection or manual quantification using structural brain MRI images based on characteristics such as the volume loss, shape variance and high intensities. However, it is a subjective process with inter-observer variance. In this paper, we propose an automated detection method for MTS based on brain MRI image analysis. It includes brain and hippocampus segmentation followed by extraction of volume, shape and CSF-ratio features from the 3D hippocampal images. Support vector machines are then used for MTS detection based on the extracted features. Experimental results show that the proposed technique provides promising performance in MTS detection.