Texture analysis of digital skin images for assessing the severity of dermatoses

Abstract

Skin diseases remain widespread among the world population, even despite the constant development of methods for their diagnosis and treatment. Non-invasive skin monitoring methods based on modern biophotonics tools play an important role in studying the pathogenesis and testing new therapies for these diseases. The application of these methods in dermatological practice makes it possible to reduce the dependence of clinical decisions on the subjective visual assessment. Such methods, especially methods for visualizing skin conditions, involve processing large amounts of data, which requires the development of standardized algorithms. Textural analysis of digital skin images is one of the promising areas here [1].
Here, we propose a method for the quantitative assessment of the clinical severity of inflammatory dermatoses based on the analysis of digital RGB images of the pathological area. It involves the extraction of the texture features to monitor the dynamics of the pathology development. The proposed approach was tested in a rat model of imiquimod-induced psoriasis-like skin inflammation by comparing the extracted texture features and the resulting composite index with an overall clinical index, which reflected morphological changes in the skin. The inflammatory model was formed for six days with simultaneous recording of the experimental skin areas using a standard digital camera under fixed imaging conditions, as well as visual assessment of the severity of inflammation based on the standard PASI scoring system [2]. In addition, histological examination was performed to confirm the formation of the psoriasis-like skin inflammation. To quantitatively characterize the images, 87 texture features were calculated, including first- and second-order ones as well as Haralick features. Based on correlation analysis, the six most informative features were selected, and a composite index was further calculated from them by the means of principal component analysis. This index corresponded to the first principal component (PC1) and demonstrated a good correlation with the cumulative PASI scores (r = −0.82). The obtained results indicate that texture analysis of standard RGB images holds great promise for objective and reproducible quantitative assessment of the severity of inflammatory dermatoses in preclinical and clinical studies.

The research was supported by the Russian Science Foundation (Project No. 22-73-10194-П).

[1] L. Wei, Q. Gan, T. Ji, “Skin Disease Recognition Method Based on Image Color and Texture Features,” Comput. Math. Methods Med., vol. 2018 (1), p. 8145713.
[2] E. Puzenat, V. Bronsard, S. Prey, et. al., “What are the best outcome measures for assessing plaque psoriasis severity? A systematic review of the literature” Dermatology Venereol., vol. 24, pp. 10-16.

Speaker

Mariia Saveleva
Saratov State University
Russia

Discussion

Ask question