Image quality issues in teledermatology: A comparative analysis of artificial intelligence solutions

Abstract

To the Editor: Deep learning artificial intelligence (AI) models capable of classifying skin lesions have become common in recent years. 1 Esteva A. Kuprel B. Novoa R.A. et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017; 542: 115-118https://doi.org/10.1038/nature21056 Crossref PubMed Scopus (54) Google Scholar , 2 Codella N.C.F. Gutman D. Celebi M.E. et al. Skin lesion analysis toward melanoma detection: a challenge at the 2017 international symposium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC). in: 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018). IEEE, 2018: 168-172 Crossref Scopus (407) Google Scholar , 3 Brinker T.J. Hekler A. Enk A.H. et al. Deep learning outperformed 136 of 157 dermatologists in a head-to-head dermoscopic melanoma image classification task. Eur J Cancer. 2019; 113: 47-54https://doi.org/10.1016/j.ejca.2019.04.001 Abstract Full Text Full Text PDF PubMed Scopus (146) Google Scholar , 4 Lopez A.R. Giro-i-Nieto X. Burdick J. Marques O. Skin lesion classification from dermoscopic images using deep learning techniques. in: 2017 13th IASTED international conference on biomedical engineering (BioMed). IEEE, 2017: 49-54 Google Scholar Many models are built using high-quality professional images from publicly available datasets, introducing biases to the data. 5 Bissoto A. Valle E. Avila S. Debiasing skin lesion datasets and models? Not so fast. in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. IEEE, 2020: 740-741 Google Scholar However, with the rise in teledermatology, patient-recorded images are often taken in poor, unnatural lighting, increasing the likelihood of inadequate contrast, color, or exposure, which presents a challenge to those models.