Deep Learning-Based MRI Analysis for Early Brain Tumor Detection

One dangerous type of cancer that arises from abnormal and uncontrolled cell division is a brain tumor. Recent developments in deep learning have proven advantageous for the health sector, particularly in the area of medical imaging for disease diagnosis. The most commonly used algorithm in machine learning for visual learning and image identification is CNN. On the other hand, our work demonstrates the use of convolutional neural networks (CNNs) for data augmentation and image processing, as well as the capacity to distinguish between malignant and non-cancerous regions in brain MRI scan images. We evaluated our scratched CNN model's performance against pre-trained VGG-16, ResNet-50, MobileNet, and EfficientB0 models using the transfer learning methodology. Despite using a fairly small dataset for the experiment, the outcomes show the efficacy and low complexity rate of our model. We achieved 94.7% for VGG-16, 96.3% for ResNet-50, 94.8% for MobileNet, and 99.3% for EfficientB0. When compared to other pre-trained models, our model performs far better in terms of accuracy with substantially less processing power usage.