Sweat Droplets Detection Using Image Segmentation on Skin Surface for Evaluation of Sweating Responses to Thermal Stimulus in Atopic Dermatitis

In this paper, we propose a method for automatically detecting sweat droplets from images taken by replicas of the silicone skin surface created by the Impression Mold Technique (IMT) to evaluate sweating responses to thermal stimulus in atopic dermatitis. The proposed method is a U-Net based image segmentation using deep learning. From experimental results, we succeeded in obtaining a result close to the expert’s manual identification result of sweat droplets.