Privacy-Preserving Chest X-ray Classification in Latent Space with Homomorphically Encrypted Neural Inference

2025-06-18

AI Summary

Homomorphic encryption (HE) incurs prohibitive computational overhead in privacy-preserving chest X-ray classification, and struggles to process high-resolution medical images directly. Method: We propose an end-to-end encrypted inference framework: (1) compress raw X-rays into a latent space using VQGAN (8× compression ratio); (2) design lightweight, low-degree polynomial approximations of activation functions and a HE-adapted CNN architecture; and (3) incorporate an enhanced Squeeze-and-Excitation module to improve feature discriminability. Results: Evaluated on two public chest X-ray datasets, our encrypted inference achieves accuracy within <1.2% of the plaintext baseline while reducing HE computation by over 70%. To our knowledge, this is the first work enabling HE-compatible latent-space classification for chest radiographs—ensuring data remains local while maintaining diagnostic accuracy and deployment efficiency, demonstrating strong clinical applicability.

Technology Category

• Machine Learning: Hardware-aware ML
• Computer Vision: Adversarial Attacks & Robustness
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• Security and Privacy: Privacy-enhancing technologies
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Medical imaging data contain sensitive patient information requiring strong privacy protection. Many analytical setups require data to be sent to a server for inference purposes. Homomorphic encryption (HE) provides a solution by allowing computations to be performed on encrypted data without revealing the original information. However, HE inference is computationally expensive, particularly for large images (e.g., chest X-rays). In this study, we propose an HE inference framework for medical images that uses VQGAN to compress images into latent representations, thereby significantly reducing the computational burden while preserving image quality. We approximate the activation functions with lower-degree polynomials to balance the accuracy and efficiency in compliance with HE requirements. We observed that a downsampling factor of eight for compression achieved an optimal balance between performance and computational cost. We further adapted the squeeze and excitation module, which is known to improve traditional CNNs, to enhance the HE framework. Our method was tested on two chest X-ray datasets for multi-label classification tasks using vanilla CNN backbones. Although HE inference remains relatively slow and introduces minor performance differences compared with unencrypted inference, our approach shows strong potential for practical use in medical images

Problem

• Privacy-preserving classification of sensitive chest X-ray images
• Reducing computational cost of homomorphic encryption for large medical images
• Balancing accuracy and efficiency in encrypted neural network inference

Innovation

• VQGAN compresses images into latent space
• Approximates activation functions with low-degree polynomials
• Adapts squeeze and excitation for HE framework