Enhanced Brain Tumor Classification using Optimized AlexNet CNN Model with Hyperparameter Tuning

Brain tumors are a leading cause of disability and mortality worldwide, making their timely and accurate detection essential for improving patient survival and quality of life. This study explores the influence of hyperparameter tuning on the performance of Convolutional Neural Networks (CNNs) in brain tumor classification using a dataset of 7023 MRI images categorized into pituitary, glioma, meningioma, and no tumor. Leveraging a tailored AlexNet architecture, the model was trained with varying epochs (10, 15, and 20), a batch size of 32, and the Adamax optimizer with a learning rate of 0.001. The optimized model achieved a training accuracy of 99.31% and an AUC of 0.9999, with a validation accuracy of 96.15% and an AUC of 0.9973. On the test set, it demonstrated a classification accuracy of 98%. Metrics such as Fmeasure, ROC, precision, recall, and accuracy underscore the robustness of this approach in enhancing brain tumor detection.