Artificial Intelligence in Teledermatology
Purpose of Review
This review summarizes current and prospective applications of artificial intelligence (AI) and smartphone technologies to automated diagnosis and teledermatology.
Healthcare institutions are rapidly scaling up telehealth programs and embracing long-distance consultation. New developments in deep learning, a type of artificial intelligence, have reached dermatologist-level performance in diagnosing cases of melanoma from lesion images. The smartphone industry projects that next-generation devices and widespread adoption will put deep learning-capable hardware in the hands of consumers everywhere in the coming decade.
The expansion in teledermatology programs over the past decade is driven by efforts to lower cost of care, expand access to underserved areas, and improve the monitoring of chronic conditions. Although long-distance diagnosis still underperforms relative to traditional, in-person diagnosis, deep learning technologies have demonstrated the potential to achieve results on par with face-to-face care. Current mobile app diagnosis systems rely on unproven technologies which do not achieve the same standard of accuracy. Over the next few years, research in teledermatology must refine deep learning methods to work with highly variable smartphone images in order to achieve functional long-distance diagnoses.
KeywordsMachine learning Automated diagnosis Artificial intelligence Dermatology Teledermatology Smartphones
Compliance with Ethics Guidelines
Conflict of Interest
Mulin Xiong, Jacob Pfau, Albert T. Young, and Maria L. Wei declare no conflict of interest.
Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.
Papers of particular interest, published recently, have been highlighted as: • Of importance
- 1.Codella NCF, Gutman D, Celebi ME 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). 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018). 2018. https://doi.org/10.1109/ISBI.2018.8363547.
- 2.Rastgoo M, Lemaître G, Morel O, et al. Classification of melanoma lesions using sparse coded features and random forests. SPIE Medical Imaging. 2016. https://doi.org/10.1117/12.2216973.
- 3.• LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521:436–44. This survey provides a layman’s introduction to deep learning and gives an overview of the widespread success of CNN architectures particularly in setting the state-of-the-art in facial recognition and object detection. CrossRefGoogle Scholar
- 6.Szegedy C, Ioffe S, Vanhoucke V, et al. Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning. Thirty-First AAAI Conference on Artificial Intelligence. 2017.Google Scholar
- 7.• Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 542:115–118 2017. This paper demonstrated dermatologist level performance, 91% AUC, on binary classification between malignant melanomas versus benign nevi using a pre-trained Google Inception v3 CNN architecture. CrossRefGoogle Scholar
- 16.• Rat C, Hild S, Rault Sérandour J, Gaultier A, Quereux G, Dreno B, et al. Use of smartphones for early detection of melanoma: systematic review. J Med Internet Res. 2018. https://doi.org/10.2196/jmir.9392. This systematic review provides an overview of currently available teledermatology, mobile-based apps and concludes that that all studies done so far relied on heavily biased patient samples, and so no conclusion regarding the efficacy of these automated-diagnoses can be drawn. CrossRefGoogle Scholar
- 18.Done N, Oh DH, Weinstock MA, Whited JD, Jackson GL, King HA, et al. VA Telederm study: protocol for a stepped-wedge cluster randomised trial to compare access to care for a mobile app versus a workstation-based store-and-forward teledermatology process. BMJ Open. 2018;8:e022218.CrossRefGoogle Scholar
- 21.American Hospital Association. Telehealth: helping hospitals deliver cost-effective care. Issue Brief. 2016;30:1–7.Google Scholar
- 22.Inova - World-Class Healthcare for Northern Virginia and the DC Metro Area. https://www.inova.org/. Accessed 5 Apr 2019.
- 23.Chous AP. Pros and cons of using an AI-based diagnosis for diabetic retinopathy. In: Optometry times. 2018. https://www.optometrytimes.com/article/pros-and-cons-using-ai-based-diagnosis-diabetic-retinopathy-0. Accessed 5 Apr 2019.
- 24.Ericsson Mobility Report November 2018 – Ericsson. In: Ericsson.com. https://www.ericsson.com/en/mobility-report/reports/november-2018. Accessed 27 Feb 2019.
- 25.AI smartphones will soon be standard, thanks to machine learning chip. In: Futurism. https://futurism.com/ai-smartphones-machine-learning-chip. Accessed 25 Feb 2019.
- 26.Bennis M. Smartphones will get even smarter with on-device machine learning. In: IEEE Spectrum: technology, engineering, and science news. https://spectrum.ieee.org/tech-talk/telecom/wireless/smartphones-will-get-even-smarter-with-ondevice-machine-learning. Accessed 25 Feb 2019.
- 28.Health C for D and R Mobile Medical Applications. https://www.fda.gov/medical-devices/digital-health/mobile-medical-applications. Accessed 25 Feb 2019.
- 29.FDA permits marketing of artificial intelligence-based device to detect certain diabetes-related eye problems. In: FDA news releases. https://www.fda.gov/newsevents/newsroom/pressannouncements/ucm604357.htm. Accessed 5 Apr 2019.
- 30.Evans data corporation | global developer population and demographic study. 2018;2. https://evansdata.com/reports/viewRelease.php?reportID=9. Accessed 25 Feb 2019.
- 33.Izadi S, Mirikharaji Z, Kawahara J, et al. Generative adversarial networks to segment skin lesions. 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018) 2018;881–4.Google Scholar
- 35.Rizzi M, D’Aloia M, Cice G. Computer aided evaluation (CAE) of morphologic changes in pigmented skin lesions. In: Murino V, Puppo E, Sona D, Cristani M, Sansone C (eds) New trends in image analysis and processing -- ICIAP 2015 workshops. Springer International Publishing; 2015. p. 250–257.Google Scholar