MRI Based Brain Tumor Detection Using Deep Learning Methods – YOLOv5

Abstract

The brain is the main nervous system in the human body. The brain provides all the instructions to the whole human body. The brain can be affected by a tumor or irregular tissues when our brain is affected by a tumor we called it a brain tumor. The brain tumor is mainly of two types first one is the Primary Tumor and the second one is the Secondary Tumor. This tumor can be a cause of human death. Brain tumors have mainly three different classes first one is Meningioma tumor second one is the Glioma tumor and the last one is the Pituitary tumor. Among these several classes of tumor Gliomas are the most dangerous and life-threatening brain tumor with exceptionally quick development. Gliomas detection using CAD (ComputerAided Detection) is a very challenging task. Because of different shapes, sizes, and boundaries of tumors with the surrounding area. Magnetic Resonance Imaging (MRI) is the most widely used method for imaging structures in the human brain. In this study, a deep learning-based method is used which has different modalities for the detection of brain tumors. The proposed YOLOv5 deep learning method is used for detecting brain tumors. YOLO stands for You Only Look Once. YOLOv5 model is mainly a branch of CNN (Convolution Neural Network) with 32 layers. This model is used because it has the power to detect small things. The proposed method is validated on the BRATS dataset. We have used this data with the process of data preprocessing, normalization, and data labeling with different classes of brain tumors. We used 80% of the data for training and 20% of the data for validation. After training the YOLOv5 model we have tested 20% of different data and got an accuracy of 91%.