Modern Diagnostic Imaging Classifications and Risk Factors for 6G-enabled Smart Health Systems

Abstract

The creation of smart healthcare systems is a viable strategy to improve the quality and availability of healthcare services. Identity theft, data breaches, and denial-of-service attacks are just some of the security concerns that have arisen as a result of connecting wireless networks and smart medical equipment. A secure and trustworthy smart healthcare system that can protect patient data and preserve the confidentiality of private medical information is especially important in light of these vulnerabilities. Medical diagnosis assumes increasing importance as the amount of data created daily in the 6G-enabled Internet-of-Medical Things (IoMT) grows exponentially. To enhance the anticipation accuracy and supply a real-time medicinal diagnosis, this research presents an approach integrated into the 6G-enabled IoMT that requires less human intervention for healthcare applications. To do this, the proposed system combines deep learning with optimization methods. MobileNetV3 architecture is then used to learn the features taken from each image. In addition, we improved the performance of the HGS-based arithmetic optimization algorithm (AOA). The operators of the HGS are used in the new approach, dubbed AOAHG, to improve the AOA operation capacity as the viable province is divided up. We design a 6G-enabled IoMT approach that requires fewer humans in healthcare settings but yields faster diagnostic results. The new approach was developed to be used in systems with limited means. The created AOAHG prioritizes the most important features and guarantees an overall upgrade in model categorization. When compared to other methodologies in the literature, the framework’s results were impressive. The created AOAHG also outperformed alternative FS methods in terms of the achieved accuracy, precision, recall, and F1-score. For instance, AOAHG had 92.12% accuracy with the ISIC dataset, 98.27% with the PH2 dataset, 95.24% with the WBC dataset, and 99.84% with the OCT dataset.