Medical Image Analysis for Brain Tumor Detection Using Convolutional Neural Networks

Brain tumors are serious medical conditions that can be life-threatening if not detected early. Traditional methods of identifying brain tumors require experienced doctors and can take a long time. This research presents an automated system that uses artificial intelligence to detect brain tumors from medical images quickly and accurately. Our study utilizes deep learning techniques, specifically two widely recognized pre-trained models: VGG16 and VGG19. We curated and labeled a dataset of brain scan images, separating them into two categories: images with tumors and images without tumors. To strengthen our system’s learning capability, we augmented the data by rotating and flipping the original images, resulting in a dataset four times larger. The models were trained to identify visual patterns indicative of tumors. Quantitative evaluation was performed to assess model performance. VGG16 achieved an accuracy of 94.5%, F1-score of 0.93, sensitivity of 95.2%, and specificity of 93.7%. Similarly, VGG19 yielded an accuracy of 95.8%, F1-score of 0.94, sensitivity of 96.1%, and specificity of 94.8%. While more recent architectures such as ResNet, DenseNet, and Vision Transformers (ViT) are available, VGG16 and VGG19 were selected in our study for their proven effectiveness in medical imaging tasks, efficient training requirements, and compatibility with our dataset size. This choice provides a robust benchmarking baseline for brain tumor detection, enabling meaningful comparison for future work. To bridge the gap between research and clinical practice, we developed a user-friendly website where doctors can securely upload brain scan images and instantly receive AI-driven analysis. The platform features user registration and secure login, ensuring patient confidentiality. By providing fast and reliable tumor detection results, our system complements radiologists' diagnosis, helping them prioritize cases, minimize human error, and make timely decisions that can potentially save lives.