Federated learning (FL) enables collaborative medical image analysis without centralising sensitive data, making it highly suitable for privacy-critical applications such as brain tumour detection from magnetic resonance imaging (MRI). However, conventional FL frameworks remain vulnerable to parameter-level information leakage through shared model updates, which may expose sensitive information about the underlying medical data. This study proposes a privacy-preserving FL framework that integrates Cheon-Kim-Kim-Song (CKKS)-based homomorphic encryption (HE) for secure aggregation with an Xception convolutional neural network (CNN) for multi-class brain tumour classification. In the proposed framework, local models are trained in plaintext on distributed clients, while the resulting model parameters are encrypted before transmission and aggregation. This approach ensures that the central server performs aggregation without gaining access to plaintext model updates, thereby improving confidentiality and reducing the risk of privacy breaches. Experiments conducted on a four-class brain MRI dataset demonstrate that the proposed approach achieves a global classification accuracy of 97.87%, with less than 1% performance degradation compared to conventional non-encrypted FL. Furthermore, a comprehensive analysis of encryption, decryption, and federated communication round latency is performed to evaluate the computational overhead introduced by CKKS-based HE. The experimental findings confirm the practical feasibility of integrating HE-assisted secure aggregation with deep learning-based FL systems for privacy-preserving medical imaging applications, while maintaining high diagnostic performance and reliable collaborative model training.
Anik Sen, Swee-Huay Heng, Shing-Chiang Tan· International Conference on...· 0 citations
Federated Learning enables collaborative model training across distributed clients without centralizing raw data, offering privacy advantages for sensitive domains such as healthcare and finance. However, recent work has shown that standard federated protocols remain vulnerable to gradient inversion attacks, where a malicious aggregator can reconstruct private training samples from shared model updates. Homomorphic Encryption provides a cryptographic solution by allowing computation on encrypted data, but existing encrypted federated systems have not been rigorously evaluated under realistic data heterogeneity, a condition where clients hold non-identically distributed data. We present PryML, a privacy-preserving federated learning framework that integrates CKKS approximate homomorphic encryption with the Flower federated learning ecosystem. We provide a formal convergence analysis proving that CKKS encryption noise contributes additively rather than multiplicatively to heterogeneity-induced convergence error, meaning encryption does not amplify the degradation caused by skewed data distributions. Experimental validation on the MNIST dataset with 10 clients over 20 communication rounds demonstrates 99.0 percent accuracy under uniform data distribution and 98.0 percent under extreme label skew with a heterogeneity index of approximately 0.69, maintaining less than 0.1 percentage point gap compared to unencrypted baselines across all tested conditions. Security evaluation confirms resistance to gradient inversion attacks, reducing reconstruction similarity from 40 percent to 2.1 percent, with ciphertext entropy reaching 7.95 bits per byte. PryML provides practical encrypted federated learning for privacy-critical applications requiring regulatory compliance.
Alareqi Mohammed Muneer Mohammed Thabit, Heng Swee Huay, Anik Sen· International Conference on...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
