Classification of Brain Tumours in Medical Images Through the Application of Machine Learning

Early detection of brain tumors is crucial, yet existing segmentation methods have not yielded satisfactory results. Brain tumors are deadlier than many other cancers, primarily because their complex structures often merge with normal tissues, complicating classification. This study explores machine learning techniques to classify brain tumors—specifically Glioma, Meningioma, and Pituitary tumors—using an MRI scan dataset. We employ convolutional neural networks (CNNs) and ensemble approaches, integrating advanced pre-processing, feature extraction, and model optimization for improved accuracy. Our models achieve state-of-the-art precision, recall, and F1 scores, outperforming traditional methods. Future work will expand the dataset and incorporate multimodal imaging and federated learning to improve generalization and privacy.