Weakly Supervised Prostate Cancer Prediction

The project entails the implementation of classification models for predicting prostate cancer, coupled with the utilization of Explainable AI techniques such as Grad-CAM to interpret model decisions. Confronting challenges such as limited supervision and class imbalance, the project employs various strategies, including the utilization of pretrained models from other datasets, transfer learning methods, and unsupervised/self-supervised pretraining. Furthermore, the project incorporates resampling techniques, data augmentation strategies, and loss adjustment strategies to enhance the robustness and performance of the models.