Brain Tumor Detection using Deep Learning: YOLO-v8 and VGG-19 in Medical Imaging

This paper introduces a comprehensive approach to brain tumor detection using MRI scans, leveraging advanced deep learning models—VGG-19 and YOLO-v8. The VGG-19 model demonstrated exceptional classification accuracy of 98 percentage, effectively distinguishing tumor regions from non-tumor areas. In comparison, YOLO-v8 achieved a slightly higher accuracy of 98.6 percentage, highlighting its strength in real-time detection, which is crucial for time-sensitive clinical applications. To enhance model performance, various preprocessing techniques were applied, including resizing, normalization, and data augmentation. Transfer learning and fine-tuning were utilized to improve learning efficiency, with pre-trained weights from large-scale datasets such as ImageNet (for VGG-19) and COCO (for YOLO-v8), reducing training time while increasing accuracy. Both models were trained on a diverse MRI dataset over 50 epochs, with hyperparameters such as learning rate and batch size carefully adjusted for optimal results. Performance evaluation was conducted using precision, recall, and F1-score to ensure a rigorous assessment of the models. The findings indicate that VGG-19 excels in classification tasks, making it highly suitable for diagnostic applications requiring high accuracy and detailed image analysis. Meanwhile, YOLO-v8’s efficiency in real-time detection enhances its applicability in clinical environments where immediate decision-making is critical. By leveraging these models, the proposed system has the potential to improve brain tumor detection accuracy and efficiency, contributing to better patient outcomes and advancements in automated medical diagnosis. Future research could explore the integration of additional deep learning architectures and expand this approach to other medical imaging modalities for broader applicability.