This project focuses on using deep learning to classify tumors in brain MRI scans. It explores different neural network architectures, including transfer learning and custom convolutional layers, to improve classification accuracy. Additionally, the Gemini 1.5 Flash model is used to generate explanations for the model's predictions, providing insights into classification decisions.
-
Train and evaluate deep learning models for brain tumor classification.
-
Use transfer learning with the Xception model to improve accuracy.
-
Develop a custom CNN model optimized for high performance.
-
Build a Streamlit web application that allows users to upload MRI scans and receive model predictions.
-
Integrate Gemini 1.5 Flash to generate explanations for the model’s predictions.
-
Compare multiple CNN models side-by-side using an interactive dashboard