Deep Learning Based Classification of Brain Tumors Using MRI

In this research endeavor, we undertook the task of classifying brain tumors utilizing Magnetic Resonance Imaging MRI data. Leveraging the ResNet50 architecture, our initial utilization of existing models, namely Long Short Term Memory LSTM and Support Vector Machine (SVM), demonstrated suboptimal accuracy for MRI classification. In response, we proposed the integration of a Convolutional Neural Network CNN to enhance model performance. Employing the ImageDataGenerator, we implemented data augmentation techniques to diversify training images and improve the model’s efficacy. The preprocessing phase involved resizing, dual filtering, and pixel value normalization. Training procedures incorporated callback mechanisms for early stopping, learning rate reduction, and model checkpointing. Subsequent evaluation on the test dataset revealed an efficient accuracy, underscoring the efficacy of the CNN approach in enhancing brain tumor classification. This methodological refinement represents a significant advancement in MRI based brain tumor classification.