Image steganography in the IoMT and telemedicine: Residual and Enhanced Discriminative Network

Through the use of technology such as computers, phones, and medical equipment, telemedicine and the Internet of Medical Things (IoMT) enable off-site interactions between medical professionals, patients, and experts. Patient privacy or consultation outcomes might be jeopardized, leading to potentially catastrophic medical events, if the information sent in IoMT is unlawfully steganographic, altered, or leaked throughout transmission and storage. There is an immediate need to safeguard the privacy and value that is the picture data included in electronic health records (EHRs). A possible option may be image steganography. Consequently, a StegaStamp-based steganographic model for pictures from medical records is created. Medical record photos are document images, not like natural photographs, and hence they are susceptible to image cropping assaults. When it comes to detecting illicitly transmitted medical pictures in IoMT and telemedicine, steganalysis plays a crucial role. To address the challenges of picture steganalysis in IoMT and telemedicine, we provide RGAN-Net, a Residual Generative Adversarial Discriminative Network. A deep residual network, a steganographic information discriminative mechanism, and an augmentation module for steganographic information make up RGAN-Net. In order to improve the illicit steganographic signal in highly detailed medical images with complex textural qualities, the RGAN-Net utilizes a steganographic information improvement module. In order to extract and compress steganographic features, a deep residual network is used. The deep residual network uses a steganographic data discriminative technique to correct the steganographic characteristics and remove high-frequency characteristics that are erroneously classified as steganographic information. The range of RGAN-Net steganalysis error PE was found to be 0.0732-0.0010 and 0.231-0.026, respectively, in experiments carried out on public and private datasets with data concealing payloads that varied from 0.1bpp/bpnzac-0.5bpp/bpnzac in the spatial realm and the JPEG domain, respectively. Overall, the RGAN-Net beats four top-tier steganography detectors, according to both quantitative and qualitative findings on both public and private datasets.