Enhanced Brain Tumor Diagnosis with EfficientNetB6: Leveraging Transfer Learning and Edge Detection Techniques

Correct identification of brain tumors is crucial for determining the subsequent steps in patient management and prognosis. This study introduces a novel approach by mimicking threeenhanced deep learning models EfficientNetB0, EfficientNetB6, and ResNet50 on a dataset of 7022 MRI instances, each depicting one of four varieties of brain tumors. The research was conducted using advanced neural network architectures, leveraging transfer learning to improve model performance. Results indicated that EfficientNetB6 achieved the highest testing accuracy at 99.39%, outperforming EfficientNetB0 and ResNet50, which recorded test accuracies of 95% and 97% respectively. Evaluation metrics further highlighted the superior performance of EfficientNetB6, with a precision, recall, and F1 score all at 99%. These findings demonstrate the significant potential of deep learning algorithms in enhancing the diagnostic accuracy of brain tumors, suggesting their implementation in clinical settings could lead to better diagnosis and treatment options.