Homomorphic-Encrypted Volume Rendering.

Area of Science:

• Computer Science
• Data Security
• Medical Imaging

Background:

• Real-time visualizations and computationally intensive tasks are often processed in data centers.
• Certain rendering applications demand stringent data confidentiality, preventing unauthorized access to sensitive information.

Purpose of the Study:

• To develop a direct volume rendering technique that operates on encrypted data, ensuring privacy.
• To enable secure, remote visualization of sensitive volumetric data without compromising confidentiality.

Main Methods:

• Implementation of direct volume rendering using the homomorphic Paillier encryption algorithm.
• Development of novel routines for encrypted-data compositing, interpolation, and opacity modulation.
• Design of a simple transfer function compatible with encrypted data processing.

Main Results:

• The proposed method renders volume data directly on encrypted data, rendering both data and images unreadable to the server.
• Performance and memory overhead analysis of the privacy-preserving scheme were conducted.
• The system guarantees data privacy even if the rendering server is compromised.

Conclusions:

• This work presents the first privacy-preserving remote volume rendering approach that does not rely on trustworthy servers.
• Data owners only need to protect their private key to ensure the confidentiality of their data and rendered images.
• The open and secure-by-design approach prevents data leakage to foreign parties, even under server compromise.