Automatic grading of Acne vulgaris using deep learning

In this study, we trained a neural network to perform automatic grading of digital images of acne patients with reliabilities comparable or superior to those of expert physicians. A dedicated device was employed to acquire images of 479 patients belonging to three different ethnic groups. A convolutional neural network trained with features extracted from local patches extracted from the facial skin showed an accuracy of 0.85 and a correlation between manual evaluation and automatized IGA of r=0.96. This is the first work where a neural network was able to directly classify acne patients according to an ordinal scale with no human intervention and no need to count lesions.