IoT enabled healthcare environment using intelligent deep learning enabled skin lesion diagnosis model

Abstract

Melanoma skin cancer is a widely occurring disease over the globe and it is needed to identify the disease at the initial stages. Recent technologies like Internet of Things (IoT) have aimed to design efficient healthcare system. Besides, deep learning and transfer learning are progressively important in the healthcare diagnoses of various diseases. This study focuses on the design of Intelligent Internet of Things with Deep learning Enabled Skin Lesion Diagnosis (IIoT-DLSLD) model for healthcare environment. The proposed IIoT-DLSLD technique enables the IoT devices to capture the dermoscopic images and classify the skin lesions into distinct class labels. In addition, the IIoT-DLSLD technique employs Gaussian filtering (GF) based preprocessing technique to eradicate the presence of noise. Besides, Otsu thresholding with swallow swarm optimization (OT-SSO) algorithm is employed for skin lesion segmentation. For skin lesion classification, MobileNet v2 based feature extraction and emperor penguin optimizer (EPO) based deep wavelet neural network (DWNN) is utilized. The experimental validation of the IIoT-DLSLD technique take place against ISIC dataset and the results portrayed the betterment of the IIoT-DLSLD algorithm over the other state of art skin lesion classifiers.

Keywords

• Healthcare
• Artificial intelligence
• Skin lesion classification
• Internet of Things
• Deep learning
• Lesion segmentation