CLASSIFICATION OF BRAIN HEMORRHAGE USING NEURAL NETWORKS AND TRANSFER LEARNING

Brain hemorrhage is a form of stroke triggered by a brain artery burst in the brain that leads to localized bleeding in the surrounding tissues. It is a severe medical condition that requires urgent treatment. Brain Hemorrhage, also known as Intracranial hemorrhage (ICH) is detected using CT (Computed Tomography) scan and MRI (Magnetic Resonance Imaging) scan. The manual interpretation of CT scans is a tedious task for radiologists. This work proposes two methods to identify brain hemorrhages by classifying the CT scan images into hemorrhage and non-hemorrhage images. One of the methods uses transfer learning while the other is by creating a CNN from scratch. Both methods use Convolutional Neural Network and Dense Neural Network to classify brain hemorrhages. This work will help doctors and radiologists in the early detection of acute brain hemorrhages which will help in the treatments of patients. The proposed CNN model achieved an accuracy of 80% in classifying the brain CT images. The VGG19 model performed the same classification with an accuracy of 95%.