Deep learning for segmentation of brain lesions from CT scans

• Main Author: Chin, Luke Peng Hao
• Other Authors: Jagath C Rajapakse
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Computer science and engineering::Computer applications::Life and medical sciences; Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: https://hdl.handle.net/10356/166225
• Institution: Nanyang Technological University

Intracranial Hemorrhage (ICH) is a brain abnormality that occurs when blood vessels rupture and acute bleeding occurs within the brain. Urgent treatment is treatment as ICH can result in hemorrhagic stroke, which is a potentially fatal and neurologically damaging condition. The most common modality of ICH diagnosis is through Computed Tomography (CT) scans, which require an experienced radiologist to analyse these scans. Hence, the aim of the project is to accelerate the diagnosis process by developing automatic deep learning models to help segment the ICH lesions produced from CT scans. A novel method is proposed, which leverages bounding boxes to help with the segmentation of ICH lesions. Our experiments showed that there were significant improvements in the segmentation results when the lesions underwent these steps compared to direct segmentation, and also provided insights for more ways to improve the segmentation process in the future.