EXPLAINABLE DEEP LEARNING FOR ENHANCED PRECISION IN BRAIN TUMOR CLASSIFICATION AND PROGRESSION FORECASTING
Keywords:
Brain Cancer, Brain Cancer Classification, Brain Cancer Detection, Machine Learning, Multimodal imagesAbstract
The project establishes a responsive web-based AI platform to detect early-stage brain tumors from MRI images, particularly in areas with limited medical facilities, such as rural areas in Pakistan. The algorithm uses MRI scans and applies machine-learning and deep-learning models (CNNs, RNNs, and decision trees) to forecast tumor behavior and inform treatment decisions. It uses explainable artificial intelligence, such as SHAP, LIME, and Grad-CAM, to generate automated medical reports. It employs Python frameworks such as TensorFlow, PyTorch, and Scikit-learn to build the platform and integrates data with MongoDB, MySQL, and DICOM/PACS systems. The system is tested using metrics such as F1-score, sensitivity, and AUC, which show better results than traditional diagnostic methods.
