Fabrication of a Novel Portable Device for Easy Detection of Neonatal Atopic Dermatitis using Deep Learning

An itchy inflammation of skin is Atopic dermatitis. It generally develops at the age lower than 5 years; but can occur to people of any age group. It is common in People who are having the medical history of the disease in their family. Hay fever and Asthma are few of the reasons of this disease. More than 10 million people suffers from this disease every year. Along-lasting disease which can occur periodically even after a cleanup. The common symptoms of this disease are Itchiness, rashes, red to brown patches on the skin's various parts, & dryness and specially the face and scalp in the case of infant. Antibiotics, creams having corticosteroids, oral drugs all are having relieving factor on short term basis; but all are having various side effects so it's harmful to use these treatments to the children. As children below 5 years cannot express their problem so it's become difficult to get what the children are facing through because of this disease. In this paper we are going to use the deep learning approach for automated diagnosis for atopic dermatitis in infant. It is an image-based approach (convolution neural networks) in which various samples have been collected. Based on the images, classification has been done to diagnose different symptoms associated with Atopic dermatitis in infant so that the early diagnose of the disease can be done and treatment are provided to the children. FIG.1: Sample pictures (Pseudocode for the proposed Atopic Dermatitis diagnosis approach) Feature Extraction & Classification using pre-trained CNN (ResNet) Atopic t of Demardr Knowledge Base Feature Output Extraction using Diagnosis of Trained ResNet Atapie Dermatitis Model Testing Image Fig. 2: Major steps of the proposed approach.