AI 2012 - LNAI 7691

Automatic Scoring of Erythema and Scaling Severity in Psoriasis Diagnosis

Juan Lu¹, Ed Kazmiercazk¹, Jonathan H. Manton², and Rodney Sinclair^3,4

¹Department of Computing and Information Systems, University of Melbourne, Australia
jualu@student.unimelb.edu.au
edmundak@unimelb.edu.au

²Department of Electrical and Electronic Engineering, University of Melbourne, Australia
jmanton@unimelb.edu.au

³Department of Medicine(Dermatology), University of Melbourne, Australia
rodneyds@unimelb.edu.au

⁴St. Vincent’s Hospital Melbourne, Australia

Abstract. Psoriasis is a common skin disease with no known cure. It is both subjective and time consuming to evaluate the severity of psoriasis lesions using manual methods. More objective automated methods are in great demand in both psoriasis research and in clinical practice. This paper presents an algorithm for scoring the severity of psoriasis lesions from 2D digital skin images. The algorithm uses the redness of the inflamed skin, or erythema, and the relative area and roughness of the flaky scaled skin, or scaling, in lesions to score lesion severity. The algorithm is validated by comparing the severity scores given by the algorithm against those given by dermatologists and against other automated severity scoring techniques.

Keywords: Skin image analysis, Severity modelling, Classification, Computer-aided diagnosis, Psoriasis

LNAI 7691, p. 73 ff.

Full article in PDF | BibTeX