<![CDATA[Brain tumors are among the most dangerous diseases that require fast and accurate detection and diagnosis. One of the most commonly used medical imaging modalities for brain tumor diagnosis is Magnetic Resonance Imaging (MRI). However, manual identification and classification of brain tumors in MRI images are time-consuming and highly dependent on the expertise of medical specialists. Therefore, this study aims to apply a transfer learning method based on the EfficientNet-B5 architecture for automatic brain tumor classification in MRI images. The dataset used in this research is the Brain Tumor MRI Dataset obtained from the Kaggle platform, which consists of four classes: glioma, meningioma, pituitary tumor, and no tumor. The research stages include image preprocessing, data augmentation, dataset splitting into training, validation, and testing sets, as well as model training using a transfer learning approach. The EfficientNet-B5 architecture is utilized as a feature extractor with the addition of custom classification layers. Model performance is evaluated using accuracy, precision, recall, F1-score, confusion matrix, and ROC-AUC metrics. The results show that the EfficientNet-B5 model achieves excellent classification performance with an accuracy of 98.75%, precision of 98%, recall of 98%, F1-score of 98%, and a ROC-AUC value of 0.99 in multi-class classification. These findings indicate that the EfficientNet-B5-based transfer learning approach is effective for classifying brain tumors in MRI images and has the potential to assist medical professionals in achieving faster and more accurate diagnostic processes.]]>
Copyrights © 2026
