<![CDATA[Brain tumors are deadly diseases with a high mortality rate, making early diagnosis crucial to improving patient survival rates. However, manual diagnosis through Magnetic Resonance Imaging (MRI) often requires significant time and is prone to errors. This study developed an MRI image classification method using the EfficientNetB3-based Convolutional Neural Network (CNN) architecture to detect brain tumors. The dataset used was obtained from Kaggle, consisting of 253 brain MRI images, including 98 normal and 155 abnormal images. The data were preprocessed through normalization and resizing to 224x224 pixels. The model employed transfer learning techniques using pretrained weights from ImageNet, enhanced with additional layers to improve performance. Evaluation was conducted using metrics such as accuracy, precision, recall, F1-score, AUC, as well as confusion matrix and classification report analyses. The results showed that the EfficientNetB3 model achieved an overall accuracy of 86%, demonstrating its capability to support brain tumor diagnosis processes quickly and accurately. This implementation is expected to provide a significant contribution to early detection of brain tumors and improve patient care quality in the medical field.]]>
