Intracranial hemorrhage detection using deep convolutional neural network

Abstract

A brain hemorrhage is a serious medical emergency that can cause intracranial bleeding that occurs inside the cranium. Intracerebral hemorrhage leads to severe neurological symptoms on one side of the human body, such as loss of consciousness, numbness, or paralysis. That often needs swift and intense therapy. Hypertension specialists review the patient's cranial medical images to see the location of intracranial bleeding. Now, it is a complex process and often time-consuming. This research identifies a convolutionary neural network approach from computed tomography scans for automatic brain hemorrhage detection. Convolutional neural networks are a powerful image-recognition technique. This research evaluates a firm neural network optimized for the detection and quantification of intraperitoneal, subdural/epidural, and subarachnoid hemorrhage on contrast CT scan. The dataset used for this research includes 180 GB images of 3D head CT studies (more than 1.5 million 2D images). All provided images are in DICOM format used for medical images.