Design and development of a deep learning model for brain abnormality detection using MRI

The research aims to develop a DL model for the detection of abnormalities in MRI images that works as an automated and accurate detection system that assists health care professionals in diagnosing the abnormalities in brain. In this research, an advanced brain abnormality prediction model associated with the deep Convolutional Neural Network (CNN) is implemented. The main advantage of this research is the proposed sonar emigration optimization that uses sonaring behaviour for predicting the position of the target with an improved convergence rate. Additionally, intensity, texture and shape-based features extract significant features for enhancing the prediction results. The sonar emigration-based deep CNN-based classifier attained the values of 95.46%, 95.72%, 94.56%, and 96.39% for dataset-1 during TP 90 for accuracy, sensitivity, specificity, and F1 score. For dataset-2 the proposed method attained the values of 94.15%,94.40%,93.25% and 95.07%, during the TP 90 while measuring the metrics, which is quite more efficient than other methods.