Detection of Intracranial Hemorrhage and Its Subtypes from CT Scan Images Using Multilevel Features Deep Neural Network

Abstract

Intracranial hemorrhage is a challenging diagnosis in medical emergencies, in which medical practitioners need quick decision to provide best medical facilities to patient. BuL. in practice accurate diagnosis is a time consuming job. A patient with Intracranial hemorhage requires immediate treatment, otherwise a delay may result in severe organ damage or death .To keep this medical emergency in mind, I developed a Convolutional Ncural Network based network called MLFDcepNet (Multi Level Features Deep Neural Network) which is trained on RSNA (Radiological Society of North America) dataset uploaded on Kaggle competition. Our model out performed on RSNA dataset and achieved an overall accuracy of about 92.32%. I validated our model on COVID-19 dataset available in 2020 and the results have been promising so far.