Enhancing Privacy in Oral Cancer Detection through Federated Learning: A Cross-Institutional Study

Oral cancer is one of the most distressing cancers, resulting in a high global health burden. Early detection is crucial for improving survival rates. However, conventional Machine Learning techniques for oral cancer detection require diverse datasets for better performance, which compromises data privacy. Federated Learning is a promising solution to these challenges, allowing collaborative model training without direct data sharing. This research aims to evaluate the effectiveness of federated learning in improving the accuracy and generalizability of oral cancer detection models compared to traditional centralized learning approaches while maintaining data privacy and compliance with healthcare regulations. The implementation was done using the FedAvg algorithm and the FedProx algorithm for model training and aggregation. The proposed Federated Learning-based framework demonstrated robust performance and offers a viable and effective approach for enhancing oral cancer detection to improve model accuracy and generalizability.