An Empirical Analysis on Detection and Recognition of Intra-Cranial Hemorrhage (ICH) using 3D Computed Tomography (CT) images

Abstract

An immediate rescue condition known as a brain stroke is primarily brought on by inadequate blood flow to the brain and the Cellular damage on a permanent scale is the result. Ischemic and hemorrhagic brain strokes are the two primary forms. The hemorrhagic form of brain stroke is brought on by internal bleeding, while ischemic stroke is brought on by a lack of blood flow. After this attack, the affected area of the brain won't work properly. Early detection is therefore crucial for more effective treatment. By combining image processing and soft computing techniques, computer-aided diagnosis is a type of non-invasive diagnostic tool that can assist in identifying life-threatening disease in its early stages. The application of convolutional neural networks for medical image analysis is constrained by current hardware constraints. The network design and input image size are heavily correlated. This is why patch or region-based approaches are frequently used for identification and classification tasks, frequently using just local contextual data during training and inference. During the time of detecting bleed locations, there occurs a difficulty in interpreting the small indicators which leads to misdiagnosis and makes the condition worse.