Brain Tumor Detection Using Various Deep Learning Algorithms

Early diagnosis of brain tumors plays an important role in a patient’s treatment and makes it easy to save his/her life. The conventional method of manually detecting brain tumors from brain magnetic resonance imaging (MRI) scans can be problematic and erroneous. This paper presents an automatic brain tumor detection and segmentation system that is built using some of the most popular deep learning-based object detection algorithms in the world. We have implemented seven different neural network-based object detection frameworks and algorithms that include YOLO V3 Pytorch, YOLO V4 Darknet, Scaled YOLO V4, YOLO V4 Tiny, YOLO V5, Faster-RCNN, Detectron2. For this study, we used the Brain-Tumor-Progression dataset taken from The Cancer Imaging Archive (TCIA). The models were trained on 641 MRI scan images taken from this dataset. After evaluating the experimental results of these models, we determined that the YOLO V5 model provided the best performance as it was able to reach a mAP@0.5 score of 95.07%. In contrast, the YOLO V3 Pytorch model provided the worst accuracy as it earned a mAP@0.5 score of 84.30%. Real-time implementations of these models can provide medical professionals with a highly efficient, automatic brain tumor diagnostic tool that will revolutionize the field of neuroscience.