Analysis of Skin Wound Images Using Digital Color Image Processing: A Preliminary Communication

Abstract

This article presents the use of digital image processing using hue, saturation, and intensity measurements as a technique for the color analysis of chronic wounds on the skin. An adaptive spline technique was used to segment the wound boundary in the images of venous leg ulcers. This technique was further used to approximate the position of venous leg ulcers. The amount of slough within the wound site was quantified using the software developed and was compared with a grading system based on visual inspection by an experienced clinician, and the results were compared by deriving Kappa (K) statistic. There was moderate agreement over all grades between the computer and clinician. At lower grades 1 and 2, there was excellent agreement. The results from this preliminary study suggest that this analytical technique has the potential to image process chronic skin wounds.

Sub-pixel Edge Detection of Cubic Spline in HSV Color Space

“…The core of color image edge detection is the color distance model, namely the color similarity model (Oduncu and Hoppe, 2004;Wang and Ma, 2004). The HSV space is more consistent with the human visual characteristics, and it is easier to distinguish similar colors than the RGB color space, so we choose the HSV space to detect image edge.…”

Objective Image-based Analysis of Leg ulcers - A Pilot Study

“…There was moderate agreement over all grades between the computer and the clinician, with excellent agreement at lower grades 1 and 2. [14] Another computer-assisted tissue classification (granulation, necrotic, and slough) scheme involved transformation of RGB wound images into HSI colour space and attained an overall accuracy of 87.61%, with highest kappa statistic value (0.793). [15] Measurement of the wound area and the wound colour, when implemented in a software system was found to allow a fully automated determination of two proposed healing indexes.…”

Combined Machine Learning With Multi-View Modeling for Robust Wound Tissue Assessment

“…However, they addressed separately the problems of wound shape capture and tissue classification. Attempts to extract automatically the wound area using colour measurements did not completely succeeded and semi-automatic methods were preferred (Oduncu et al, 2004). Furthermore, the results obtained on several colour spaces by direct classification on the pixels were still not acceptable, even when combining several colour and texture parameters to describe the tissues (Kolesnik and Fexa, 2004).…”