Literature Review for Automatic Detection and Classification of Intracranial Brain Hemorrhage Using Computed Tomography Scans

Intracranial Brain Hemorrhage is a serious threat to health and life; it requires immediate and efficient medical treatments. It comprises five broad categories, namely epidural hemorrhage, subdural hemorrhage, subarachnoid hemorrhage, intraventricular hemorrhage, and intraparenchymal hemorrhage. We can distinguish between these subtypes on the basis of the character of bleeding and its location in the brain region. Developments in the field of Artificial Intelligence and Machine Learning particularly Computer Vision over years help the research community to propose studies, which can be used to fight various medical diseases and emergencies. Computed Tomography scans of the brain play a significant role and are popularly used for the evaluation of intracranial hemorrhage. Location of hemorrhage on unenhanced computed tomography scans of the brain and differences in X-ray attenuation helps in detecting different subtypes of intracranial brain hemorrhage. In this article, we have provided an extensive literature review for the problem of detection and classification of intracranial brain hemorrhage in the past 15 years. We have explored the objectives and applications of the existing studies, the methods adopted for diagnosis in them, and different pre-processing techniques that were applied to image data. We concluded our study by stating some major research challenges on the basis of previously done work in the field, their possible solutions which can be followed in future works and limitations of this study. This paper aims to help and facilitate radiologists, medical experts, and other researchers in understanding the way how machine learning can be potentially used in the diagnosis of hemorrhage.