Corpus callosum index calculation and segmentation with deep learning

In this study, a system with capability of calculating corpus callosum index automatically is developed. Deep learning techniques are used in the solution that aims to be an assistant recommender system to radiology specialists. Main goal of the provided system is to make a positive effect on time management of radiology specialists by automating CCI calculation at detection and control of MS, Alzheimer, schizophrenia, dyslexia, and epilepsy. First step for the study was training the selected deep learning architecture. Training dataset is gathered from the Radiology department of the Akdeniz University. Segmentation pipeline built with trained model. Input for the segmentation pipeline is brain MR image. For the CCI calculation, first step is to detect corpus callosum with trained U-Net model. After the segmentation process, an algorithm is used for finding the thickness and length values of detected corpus callosum. With the obtained values, corpus callosum index is calculated. As a result of the thesis work CCI distribution with respect to MS sub-types is obtained.