A Cascaded Deep Network for Abnormal Region Extraction from MR Brain Images

Abstract

Medical images are the pictures of distributions of physical attributes of human organs captured by an image acquisition system. Brain tumors are cells in the brain that grow abnormally hence to understand the structure of dysfunctionality extraction of these regions are of utmost important. Abnormal region extraction through segmentation process of image processing is a primary step in assessing the extent of disease. These tumors have irregular shape and can be spatially located anywhere in the brain, which makes it a challenging task to segment them accurately enough for clinical purposes. In this paper cascaded deep convolutional neural networks were employed to extract the regions effectively. The first network focuses on extraction of brain region while the other is trained to extract the abnormal regions. Experiments were conducted on both standard and real time data samples and found that this network structure has attained considerable improvement correct detection however in some cases it yields high OSE (over segmentation error).