Abstract
Traumatic Brain Injury (TBI) disrupts the brain’s usual functioning and can lead to temporary or permanent neurological defects. Detecting and treating TBI at an early stage can considerably improve the recovery time and avoid serious complications. Doctors rely on medical imaging to diagnose TBIs. As against the manual detection methods which may overlook subtle patterns resulting in inaccuracy and inconsistency, the computational methods can continuously adapt based on new data which improves the prediction model’s accuracy. Specifically, deep learning techniques are capable of extracting useful features from unstructured data such as medical images, without any need ad manual feature engineering However, the existing review papers discuss TBI detection using DL methods only in addition to the traditional and ML methods and fail to cover the vast variety of the recent DL algorithms. This review exclusively focuses on DL algorithms for TBI detection and encompasses cutting-edge DL models used in this domain. The choicest collection of articles unveils the potential of deep neural networks to process different types of inputs including numerical data EEG, CT and MRI scan images. This comprehensive overview offers a one-stop solution for the various research interest groups to get an understanding of the different techniques used and acquire valuable insights to conduct research in different disciplines ranging from image processing to advanced deep neural networks.
Keywords: Artificial Neural Network (ANN), Brain Hemorrhage Classification based on Neural Network (BHCNet), Convolution Neural Network (CNN), Deep Learning (DL), Multi-Layer Perceptron (MLP), Recurrent Neural Network (RNN).