Automated Technique for Brain Tumor Detection From Magnetic Resonance Imaging Based on Local Features, Ensemble Classification, and YOLOv3.

In this article, the researcher explores an automated approach for detecting a brain tumor using MRI scans of the brain. In underdeveloped countries, many people are dying due to the slow detection process and other negligence of radiologists. People suffer from these diseases due to the slow process of recognition. Since the number of patients is greater than that of radiologists, there is the possibility of human error, which can cause serious damage. The detection of tumors from magnetic resonance imaging (MRI) data is an important manual task, specifically in terms of the time that the radiologist performs. In this study, the researchers sought to study state-of-the-art techniques to detect normal brain and brain tumors from MRI using machine learning techniques. The main objective of this study is to develop a novel automated technique for brain tumor detection. Through the worldwide consideration of practical literature, it is clear that traditional approaches are insufficient to resolve all uncertainties and problems. Therefore, a novel approach to examining MRI must be adapted. This study proposes two different novel techniques: one that uses ensemble classification and the other that makes use of the deep learning model of YOLOv3. In ensemble classification, two classification algorithms are used which are support vector machine (SVM) and K-nearest neighbors (KNNs). The YOLOv3 model is used to detect and outline tumor locations in the images. This study used an open-source dataset and data collected from hospitals in Lahore, Pakistan. The ensemble classifier achieved an overall accuracy of 80.50%, while the YOLOv3 model achieved higher performance with 97.80% accuracy, 97.40% precision, 98.18% recall, and a mean intersection over union (IoU) score of 0.65. These results confirm that YOLOv3 is a useful technique for identifying brain tumors.