P11 Objective measurement of skin tone using artificial intelligence

Clicks: 1
ID: 318356
2026
Free PDF
Article Quality & Performance Metrics
Overall Quality
Not rated
Combines reader engagement with the AI quality analysis. This article has not been analysed, so there is no overall score — reader engagement is measured and shown alongside.
AI Quality Assessment
Not analyzed
Readership in this journal

Ranked #216 of 302 articles by views in the british journal of dermatology

Most read Least read

Bar heights use a square-root scale. Only the 120 most-read articles are drawn; the journal has 302 in total.

Mint this article as an NFT
Not yet minted

Create a permanent, verifiable on-chain record of this article on the Scimatic Network. The NFT is held in your Journament account, and you can withdraw it to your own wallet at any time.

5 SUSD one-off · no wallet required
Sign in to mint
Abstract
Abstract Introduction and aims There is no universally accepted gold-standard for categorizing skin tone. Subjective scales, such as the Fitzpatrick phototype, do not sufficiently represent skin colour diversity, and the utility of skin colour charts is limited by metamerism. Thus, there is a need to develop an objective, reproducible method for the accurate quantification of skin tone. We use artificial intelligence to quantify skin tone, based on spectrophotometric measurements, from photographs. Methods A Konica Minolta CM-600d spectrophotometer was used to measure skin tone in the CIE Lab* colour space, enabling calculation of the individual typology angle (ITA). Measurements were obtained from four anatomic regions. At least 16 sites were assessed for all participants, and a photograph was taken at each corresponding site. A convolutional neural network (CNN) was trained to predict ITA from a previously unseen photograph and assign it to a skin tone group, with the corresponding spectrophotometric value and ITA classification used as the ground truth for comparison. Results A total of 2419 spectrophotometric measurements and corresponding photographs were obtained from 79 individuals. Data from 69 individuals (2139 photographs and spectrophotometric measurements) were used to train the CNN, with data from the remaining 10 participants reserved for testing the trained network. The model achieved an overall classification accuracy of 55%; of the 153 images misclassified, 147 (96%) were assigned to an adjacent skin tone category. The mean absolute error, root mean square error and coefficient of determination for neural network-predicted ITA vs. ground truth ITA were 11.97, 9.03 and 0.85, respectively. Conclusions A CNN can be used to categorize photographic images according to ITA, based on in vivo spectrophotometric measurements. This method represents an easy to use, objective and reproducible approach for skin tone classification.
Reference Key
openalex_W7165640223 Use this key to autocite in the manuscript while using SciMatic Manuscript Manager or Thesis Manager
Authors J SCOTT, Maria Charalambides, B Mills, Eugene Healy
Journal the british journal of dermatology
Year 2026
DOI
10.1093/bjd/ljag151.052
URL
https://doi.org/10.1093/bjd/ljag151.052
Keywords Keywords not found

Citations

No citations found. To add a citation, contact the admin at info@scimatic.org

Comments

No comments yet. Be the first to comment on this article.