Deep Learning-Based Brain Tumor Detection and Classification of MRI Images

Brain tumors are a common cause of death, ranking 10th among various causes. They can affect both adults and children, characterized by abnormal cell growth within the brain. These tumors can be classified based on location, behavior, and morphology. Accurate classification is crucial for effective patient management and treatment, as misclassification can have severe consequences. Magnetic resonance imaging (MRI) is commonly used for tumor detection, and automated methods using convolutional neural networks (CNN) have emerged in recent years. In our research, we propose a CNN-based approach for classifying brain tumors using MRI scans. Our focus is on the multi-class categorization of glioma, pituitary, meningioma tumors, and the absence of tumors. We train our CNN model from scratch and evaluate it with K-fold cross-validation for improved accuracy. Evaluation is performed on a publicly available dataset from Kaggle, comprising MRI scans with the four classes. Our proposed method achieves a 94.78% accuracy in multi-class classification, demonstrating the effectiveness of CNN for brain tumor classification. This approach aids doctors and radiologists in making accurate diagnoses, ultimately improving patient outcomes.