Hybrid Encryption Scheme for Medical Imaging Using AutoEncoder and Advanced Encryption Standard

Recently, medical image encryption has gained special attention due to the nature and sensitivity of data lack effective using innovative techniques. Several schemes have been recommended developed in an attempt improve encryption. The majority these studies rely on conventional However, such improvements come with increased computational complexity slower processing for decryption processes. Alternatively, engagement intelligent models as deep learning along exhibited more outcomes, especially when used digital images. This paper aims reduce change transferred between interested parties overcome problem building negative conclusions from encrypted In order do so, target was transfer domain encrypting features image, which are extracted float number values. Therefore, we propose a learning-based scheme autoencoder (AE) technique advanced standard (AES). Specifically, proposed is supposed encrypt digest prepared by encoder model side. On side, analogous decoder auto-decoder after decrypting carried data. enhance quality corrupted images different types noise. addition, investigated scores structure similarity (SSIM) mean square error (MSE) applying four noise: salt pepper, speckle, Poisson, Gaussian. It noticed that all noise added, reduced this resulting Finally, performance evaluation demonstrated our system improved encryption/decryption overhead 50–75% over other existing models.