Brain Stroke Prediction from Computed Tomography Images Using Efficientnet-B0

Brain stroke disease is the second-most common cause of mortality and suffering worldwide in terms of key international cause of death according to World Health Organization (WHO). Deep learning (DL) provides fast and accurate prediction results, and it has developed into a potent tool in healthcare environments for providing stroke patients with individualized clinical care. Although there is a real need for research, some research areas are not receiving enough attention for scientific inquiry despite the increasing use of DL techniques in medical treatments. DL enabled methods are frequently used together with artificial intelligence to automatically identify strokes in a non-intrusive method. But it is apparent from the study's results that in order to increase prediction performance, training and classifying quality must be improved. 2515 brain CT scan images are trained, validated, and tested using a deep learning method for brain stroke prediction. In this study, the EfficientN etBO CNN model, a deep learning framework that takes sigmoid classification into consideration. The best models with an accuracy rate of 98.72 % are selected for the CNN model from a variety of brain stroke estimation techniques.