Robust Deep Learning Approach for Brain Tumor Classification and Detection

The human brain is the humanoid system's primary controller. A brain tumor is a consequence of abnormal cell growth and division in the brain, and the progression of brain tumors leads to brain cancer. The introduction of new technology into health care can be seen as a means of improving human judgment in diagnosis. The use of computer vision in diagnostics might reduce human judgment errors. Magnetic resonance imaging (MRI) is the most dependable and secure method among the numerous medical imaging technologies. The goal of this project is to employ a robust approach that specifically identifies minor abnormalities in MRIs and predicts the presence of a tumor with high accuracy for brain cancer detection. In this paper, we demonstrate the use of deep learning techniques through convolutional neural networks (CNN) for reliable tumor detection, followed by image segmentation through a marker-based watershed segmentation algorithm to view the tumor zone. A user-friendly GUI is built to assist medical professionals in achieving the above objectives without any complications. This system can be used as a support tool by doctors and radiologists for early detection of tumors. The results can be used as a second opinion and reviewed using several performance-assessed parameters to determine whether the patient has a brain tumor or not.