Development and validation of a semiautomatic image analysis system for measuring skin desquamation with D-Squames.

Background: D-Squames, have gained wide acceptance for assessing skin desquamation. The amount of corneocytes adhering to D-Squames can be assessed visually by trained observers or by computerized image analysis. Different image analysis algorythms for the evaluation of D-Squames have been published but have not been compared with each other. It was our aim to develop an image analysis system that does not require an expensive image analysis programming tool but should be optimized for routine tasks of analysing large numbers of samples. A second objective of this study was to compare two published image analysis algorythms and visual grading.
Material and Methods: The hardware components of the system are a CCD camera connected to a frame grabber card and a light box equipped with fluorescent tubes on two sides that provide a relatively cool, diffuse and even illumination of the sample. The following features were included into the software: generation and identification of bar codes for sample identification; semiautomatic recognition of ROI (region of interest), integration of study design into the analysing process, rapid calculation of desquamation index (DI: integration of the per cent area covered by scales and their thickness distribution) and/or scaling index (SI: distribution of grey values), data storage and export for further analysis. In a first step the system was validated by examining D-Squames covering a wide range of desquamation, by examining different ROI shapes (circle and square), by performing repeat measurements with different positions of the samples and by repeat measurements after re-callibrating the system. In a second step the effect of treatment with different moisturizers was evaluated by the two image analysis parameters DI and SI and compared with hydration measurements (Corneometer).
Results: The shape of the ROI showed no influence on the results (variability < 5%). Reproducibility of measurements was satisfactory (COV CDI): 1.7%, COV (SI): 2.6%). There was a good correlation between image analysis results and visual evaluation (means of 3 technicians) (r = 0.986) as well as between the two different image analysis parameters DI and SI (r = 0.971). In the clinical study moisturizer treatment resulted in variable reduction of desquamation that was closely correlated with increase in stratum corneum hydration (r = 0.97).
Conclusion: Analysing D-Squames with the image analysis system proved to be reproducible, independent of the shape of ROI, cost effective and fast and easy to operate. It has shown to be a suitable and reliable method for the objective determination of desquamation levels.