Abstract
Brain tumors remain a critical challenge in the medical domain due to their heterogeneous nature and complex imaging characteristics. This paper explores an advanced framework combining transfer learning with EfficientNet and YOLO for precise tumor classification and segmentation. The EfficientNet-B0 model is fine-tuned using transfer learning to classify tumors into four categories: glioma, meningioma, pituitary, and no tumor. YOLOv8 is employed for high-speed, real-time segmentation. Transfer learning enhances model generalization by leveraging pre-trained weights, reducing training time, and improving performance on limited medical imaging datasets. Extensive experiments demonstrate superior performance in terms of accuracy, computational efficiency, and practical applications. This framework addresses the critical need for early and accurate tumor diagnosis, reducing the workload on medical professionals and improving treatment outcomes.
