Automated Brain Cancer Detection from MRI Scans using Convolutional Neural Networks: A Deep Learning Approach

Convolutional Neural Networks (CNNs) have significantly advanced brain tumor detection in MRI images. This study presents a CNN framework using a clinical dataset of 392 tumor images and 408 healthy images, classified into two categories: tumor and healthy. To enhance model robustness, data augmentation techniques such as rotation, zooming, and horizontal flipping were applied. The CNN architecture consists of four convolutional blocks with batch normalization and max pooling, followed by fully connected layers optimized using the Adam optimizer. Trained over 80 epochs, the model achieved a training accuracy of 93.75% and a validation accuracy of 93.55%. Precision and recall metrics indicated very high accuracy, particularly for tumor detection, with an F1-score of 0.97 for the tumor class and a precision of 1.0. Although some oscillations in validation accuracy and loss suggest potential generalization issues, the results highlight a highly promising CNN approach for automated brain cancer detection, potentially aiding in diagnosis and early treatment planning.