Proposal Medical Image Protection System Based on Hybrid CNN-Transformer Model and Chaotic Maps

Medical image encryption is a crucial area of research that focuses on protecting sensitive medical images from unauthorized access and ensuring patient confidentiality. Protecting medical images from data breaches and illegal usage is essential, given the growing use of digital imaging technology (such as magnetic resonance imaging, CT scans, X-rays, and ultrasound). In order to comply with laws such as the General Data Protection Regulation (GDPR) and the Health Insurance Portability and Accountability Act (HIPAA), medical photographs must be protected since they include sensitive information about a patient’s health. This paper introduces a proposal to ensure that medical images are confidential during transmission or storage by encryption and intends to preserve their integrity for accurate diagnosis and treatment by hashing these images through constructing a feature map image. A hybrid model of CNN and transformer is suggested in this proposed system, which is used to extract a unique image’s hash of the original medical image, presenting a feature map of the image to guarantee the integrity. In the next step, the original and hashed images encrypted depending on the using of several chaotic maps, where the hash image is encrypted by making use of a logistic map, and the RGB channels of original image is encrypted separately by using a Henon map, an Arnold map, and the Lorenz system, and finally the complete encrypted image encrypts by a 9D chaotic map. This method greatly enhanced the security and complexity of the encryption process, making it very difficult for an attacker to access or recreate the original data. The combination of these chaotic maps guaranteed a higher level of secrecy and robustness in the encryption mechanism. Additionally, this approach effectively preserves time in diagnosing the medical image since the feature map is extracted on the client side. The experimental results show this approach’s ability to effectively maintain time in diagnosing the medical image since the feature map is extracted on the client side. This proposal successfully and safely protects medical images from various attacks on the client side, channels, and servers.

Abstract licence: CC BY-NC-ND

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.