Fully Automated Segmentation of Brain Stroke Lesions Using Mask Region-Based Convolutional Neural Network

Abstract

Stroke is one of the widespread neurological diseases that occur due to abnormal blood flow in the brain. In addition, stroke has been leading the cause of death worldwide in recent years. It is very important to make a rapid and precise diagnosis of stroke disease and to determine the boundaries of stroke lesions correctly in order to increase the survival rate of patients. Computer-assisted automatic segmentation methods can be used as assistant tool in the decision phase for accurate, precise, and rapid diagnosis of stroke by physicians. In this chapter, we propose a fully automated method based on mask region-based convolutional neural network (Mask R-CNN) for segmentation of brain stroke lesions using MR images. ATLAS v2.0 publicly available stroke dataset comprising of T1-weighted MR scans is used in the chapter. Within the scope of the experimental studies in the dataset, automatic segmentation of stroke lesions using the proposed Mask R-CNN method is achieved with a dice similarity coefficient of 78.50%. The findings obtained in the study show that the proposed Mask R-CNN method can be utilized as assistant tool for segmentation of stroke lesions with accurate boundaries. © 2023 selection and editorial matter, Jyotismita Chaki; individual chapters, the contributors.