DeepKeyGen: A Deep Learning-Based Stream Cipher Generator for Medical Image Encryption and Decryption.

Area of Science:

• Medical Imaging
• Cryptography
• Artificial Intelligence

Background:

• Patient privacy in medical imaging is a growing concern.
• Secure encryption methods are crucial for protecting sensitive medical data.
• Existing encryption methods may lack efficiency or robust security for medical images.

Purpose of the Study:

• To propose a novel deep learning-based key generation network (DeepKeyGen) for medical image encryption.
• To develop a secure and efficient method for generating private keys for encrypting and decrypting medical images.
• To enhance the privacy and security of patient medical imaging data.

Main Methods:

• A deep learning-based key generation network (DeepKeyGen) was developed.
• Generative Adversarial Network (GAN) was utilized as the learning network for key generation.
• A transformation domain was designed to guide the GAN in learning the image-to-key mapping.

Main Results:

• DeepKeyGen successfully generated private keys for medical image encryption and decryption.
• Evaluation across three datasets (chest X-ray, brachial plexus ultrasound, BraTS18) demonstrated high-level security.
• Security analysis confirmed the robustness of the proposed key generation network.

Conclusions:

• DeepKeyGen offers a promising solution for secure medical image encryption.
• The proposed method effectively safeguards patient privacy by securing medical imaging data.
• Deep learning, specifically GANs, can be effectively applied to cryptographic key generation for medical applications.