Deep Learning Techniques Using Lightweight Cryptography for IoT Based E-Healthcare System
Abstract
The most recent development in the digital world is the field of electronic healthcare systems, which are used for illness modeling and prediction, evidence-based treatment, remote health monitoring, and other purposes. In this paper, skin cancer is by far the most prevalent type of cancer, particularly in the region of North America, and is a serious public health concern. The most frequent causes of skin issues are fungus, germs, allergies, and viruses, among others. Machine learning and deep learning have become more common in medicine during the past few years. Many image-based categorization systems have been developed in the medical area as a result of imaging, and these systems work well. In contrast to the common methods for image processing categorization. In this paper, we suggest a well-liked deep learning (DL) approach for the categorization of skin lesions. We choose transfer learning approaches along with a number of well-liked DL architectures, such as VGG16 and InceptionV3. Our suggested techniques were trained, validated, and tested using the HAM10000 dataset, which contains 10015 dermoscopic pictures of seven distinct skin lesions. Test accuracy was 80.42% and 84.79%, respectively.