Characterizing scars in the cerebral cortex by analyzing intensities in T2/MRI sequences

Abstract

The detection of scars in the cerebral cortex usually involves a manual process performed by radiologists, who have to face multiple troubles. For example, the bad calibration of the equipment used to get images of the cerebral cortex can cause spacial and geometrical distortions in the MRI (Magnetic Resonance Imaging) sequences. Owing to the advances in algorithms capable of analyzing MRI sequences, it is possible to automatically detect scars in the cerebral cortex in a successful way. In addition, an automatic process for finding scars can decrease the subjec tivity of human interpretations and serve as a tool to support diagnoses of diseases. In this paper, we propose a new methodology to detect scars in the cerebral cortex by means of the analysis of intensities in T2/MRI sequences. In particular, we implement three main algorithms: the region growing, thresholds, and artificial neural networks.