Skin Cancer Diagnostics via Smartphone

Around 6.3 billion smartphone users are expected in about four years (1). What would be if these devices could also be used to recognize or even exclude skin cancer very early and quickly?

Students in a research group led by Sebastian Thrun from the Stanford Artificial Intelligence Laboratory were inspired by this question. They further developed an image recognition program, successfully tested by Google to recognize dogs and cats, by training the underlying algorithm to distinguish interactively benign from malignant skin lesions. Approximately 130,000 medical images, mainly provided by the University of Edinburgh and the International Skin Imaging Collaboration, formed the starting material for the learning processes of artificial intelligence.

At the end of this training, the intelligent image recognition program was able to correctly diagnose skin cancer on approximately 370 pictures. 21 renowned dermatology experts confirmed the performance. 2021, this program will not be available to all smartphone users, because a long time detailed work is still required like to adapt the smartphone phototechnology to the algorithm. However, a decisive step towards teledermatology is done.

Read and see more
(1) Information from the Cerwall, P. & Report, E. M. Ericssons mobility report https://www.ericsson.com/res/docs/2016/ericsson-mobility-report-2016.pdf (2016)
(2) A. Esteva et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature 542, 115–118. (02 February 2017) doi: 10.1038/nature21056 please copy the link: http://www.nature.com/nature/journal/v542/n7639/full/nature21056.html