Abstract:Abstract—Brain hemorrhages is a critical medical disorder thatneeds to be diagnosed as soon as possible in order to allow for immediate medical attention. In the current study, we provide amachine approach to learning to detect various types of brain hemorrhage including subdural, epidural, and intracerebralhemorrhage using a Convolutional Neural Network (CNN) based ResNet-100 framework. We train and evaluate our CNN model usinga diverse data set of brain scans of normal cases and cases of stroke types. The application of the ResNet-100 algorithm comes from itsability to handle deep learning tasks and capture complex features inmedical images efficiently. Through extensive training andvalidation, our model achieves an impressive 95% accuracy inclassifying brain scans as normal or indicative of subdural, epidural,or intracerebral hemorrhage The high accuracy shows how wemachine learning approach is effective for multiclass brainhemorrhage detection. The proposed system has the potential to helpradiologists and healthcare professionals to accurately and effectivelyidentify various types of brain hemorrhage from medical imagesUsing the capabilities of CNN and ResNet-100, our method enables early detection and timely intervention, improving patient outcomes. Keywords: Brain hemorrhage detection, CNN, ResNet-100, subduralhemorrhage, epidural hemorrhage, Intracerebral hemorrhage, accuracy

Brain Hemorrhage Detection Using AI