How can computer vision aid in melanoma detection?

How can computer vision aid in melanoma detection? A study recently published in the Journal of the American Academy of Dermatology compared the diagnostic accuracy of computer algorithms to dermatologists using dermoscopic images.

The study involved 100 randomly selected dermoscopic images comprising of 50 melanomas, 44 naevi, and six lentigines. Researchers used both non-learned and machine learning methods to combine individual automated predictions into “fusion” algorithms. In a companion study, eight dermatologists classified the lesions in the 100 images as either benign or malignant.

Dermatologist classification had an average sensitivity and specificity values of 82 per cent and 59 per cent, respectively. Compared with a dermatologist specificity value of 59 per cent at 82 per cent sensitivity, the specificity of the top challenge algorithm did not differ significantly at 62 per cent, but the specificity of the best-performing fusion algorithm was significantly greater at 76 per cent.


The study concluded that deep learning computer vision systems classified melanoma dermoscopy images with accuracy that exceeded some but not all dermatologists.


Marchetti, Michael A. et al. (February 2018.) Results of the 2016 International Skin Imaging Collaboration International Symposium on Biomedical Imaging challenge: Comparison of the accuracy of computer algorithms to dermatologists for the diagnosis of melanoma from dermoscopic images. Journal of the American Academy of Dermatology. Volume 78. Issue 2. Pages 270 – 277.e1.

Learn more about skin cancer medicine in primary care at the next Skin Cancer Certificate Courses:

Skin Cancer Certificate Courses in Australia

Leave a Reply

Your email address will not be published. Required fields are marked *