Secure medical image transmission using deep neural network in e-health applications

Area of Science:

• Medical Imaging
• Cybersecurity
• Artificial Intelligence

Background:

• Medical image diagnosis is crucial, necessitating robust security for data transmission.
• Unauthorized access to medical images poses significant risks.
• Encryption, particularly confusion and diffusion, is vital for securing image data.

Purpose of the Study:

• To develop an advanced encryption algorithm for medical images.
• To enhance the security of medical image data transmission.
• To integrate deep neural networks with chaotic systems for improved encryption.

Main Methods:

• A deep neural network algorithm was adapted to identify critical weights impacting image encryption.
• A chaotic system was generated using random numbers derived from the deep neural network.
• Image regions of interest were identified and processed using the deep neural network for pixel redistribution and chaotic system integration.
• Both color and grayscale medical images were utilized.

Main Results:

• The proposed method demonstrated effective encryption by leveraging deep neural network weights and a chaotic system.
• The algorithm successfully generated random numbers for chaotic system creation, including row/column replacement and pixel distribution.
• Evaluation and comparison with existing methods confirmed the proposed algorithm's efficacy and worth.

Conclusions:

• The integration of deep neural networks and chaotic systems offers a powerful approach to medical image encryption.
• The method provides enhanced security for sensitive medical data during network transmission.
• This technique addresses the critical need for secure medical image handling in modern healthcare.