Internet of Medical Things (IoMT) Enabled Skin Lesion Detection and Classification Using Optimal Segmentation and Restricted Boltzmann Machines

Abstract

In recent times, Internet of Medical Things (IoMT) and cloud enabled healthcare applications and services finds helpful for effective decision-making. Melanoma is the serious kind of skin cancer, results to high death rate. Earlier identification of skin cancer can leads to maximum survival rate. But the diagnosis process becomes difficult and expensive because of the need of medical experts and complex medical equipments. To overcome this issue, the latest developments in IoMT based decision making system with maximum performance can be used. This study introduces a new IoMT based skin lesion detection and classification model using Optimal Segmentation and Restricted Boltzmann Machines (RBM), named OS-RBM model. The proposed OS-RBM model involves a series of steps namely image acquisition, gaussian filtering (GF) based preprocessing, segmentation, feature extraction, and classification. Then, optimal segmentation using artificial bee colony (ABC) with kapur’s thresholding takes place. Besides, histogram and texture feature extraction will be carried out. Finally, RBM is applied as a classifier to detect and classify the existence of skin lesion in the dermoscopic images. A detailed simulation analysis takes place for ensuring the better outcome of the OS-RBM model and the results are assessed under diverse performance measures. The experimental outcome ensured the effective classification performance of the OS-RBM model with the maximum sensitivity of 96.43%, specificity of 97.95% and accuracy of 95.68%.