Efficient intracranial haemorrhage Detection and classification in CT Images using deep learning Techniques

Abstract

Intracranial Haemorrhage (ICH) is a medical condition that is characterized by the extra vascular accumulation of blood within different intracranial spaces. It could be caused by various reasons ranging from trauma, vascular disease to poor congenital development. It is considered critical that may lead to severe disability or even death. According to the bleeding location, ICH can be further classified as Epidural Haemorrhage (EDH), Subdural Haemorrhage (SDH), Subarachnoid Haemorrhage (SAH), Cerebral Parenchymal Haemorrhage (CPH), and Intraventricular haemorrhage (IVH). The early intervention through diagnosis and appropriate treatment is utmost important to prevent further worsening of the health condition of the patient. Diagnosing a brain haemorrhage can be difficult as some people do not show any physical signs. Present diagnostic methods like cerebral angiography is complex, expensive and time consuming. Among them, computed tomography (CT) scan is preferred in emergencies because of its advantages of being cost effective, faster acquisition, and resulting in fewer contraindications. Hence, diagnosis of ICH using a non-invasive approach and early intervention in deciding the appropriate treatment is important to save lives. This should be possible even in places where there is no immediate availability of physicians for the diagnosis of ICH. newlineThis thesis is directed towards the path to design a computer aided system that can be integrated with CT systems that supports automatic detection of brain haemorrhage from medical images based on deep learning techniques. This would be valuable for queue management in a busy trauma care setting, and most importantly to facilitate the decision-making in remote locations without an immediate radiologist availability. newline