Using smartphone and deep learning technology to help diagnose skin cancer. (28th February 2020)

• Record Type: Journal Article
• Authors: Jiang, Y.Q.; Xiong, J.H.; Li, H.Y.; Yang, X.H.; Yu, W.T.; Gao, M.; Zhao, X.; Ma, Y.P.; Zhang, W.; Guan, Y.F.; Gu, H.; Sun, J.F.
• Is Part Of: British journal of dermatology. Volume 182:Number 3(2020)
• Page Start: e95
• Page End: e95
• Publication Date: 2020-02-28
• Subjects: Dermatology -- Periodicals; Skin -- Diseases -- Periodicals; 616.5
• DOI: 10.1111/bjd.18826
• Languages: English
• ISSNs: 0007-0963

Abstract

Summary: Basal cell carcinoma (BCC) is the most common skin cancer with a rapidly rising incidence. The diagnosis of BCC requires careful inspection of microscopic skin images, which is labour‐intensive due to the large number of images to be analyzed. Computer‐aided diagnosis of diseases has been developed to help the analysis of microscopic images in pathology (which is the study and diagnosis of disease by examining tissue that has been removed.) In particular, deep learning approaches, which are ways in which a machine can 'learn' to perform a task, are able to identify and capture patterns in images and have shown significantly improved performance on pathological image‐related tasks. In this study, from China, the authors aimed to develop deep learning frameworks for automatic recognition of BCC based on smartphone‐captured microscopic ocular images (MOI) instead of whole slide images (WSIs). MOIs are photographed from microscope eyepieces using smartphone cameras. A total of 8046 MOIs and 128 WSIs were collected and used to build models. Two deep learning frameworks for recognizing BCC pathologically were developed with high sensitivity and specificity (meaning accuracy). Compared with the model trained on WSI, recognizing BCC through smartphones could be considered as a future clinical choice. This is a summary of the study: Recognizing basal cell carcinoma on smartphone‐captured digital histopathology images with a deep neural network Abstract : Linked Article: Jiang et al. Br J Dermatol 2020; 182 :754–762