Inference of pressure ulcer stage and texture on an image training set

Abstract

Health assistance depends essentially on effective information providing for professionals. Presently visual information plays an outstanding role concerning patients’ status without necessity of invading procedures. The diagnosis related to stage and texture in pressure ulcers (PU) is considered as a classification procedure, based on the image colors. As a training set this study captured 18 PU images from nine persons. RGB mean values were adopted as features to characterize the images. Classification used an ID3 similar algorithm. It achieved 83.3% accuracy concerning color and stage, as well as 88.9% concerning color and texture. So the technique was effective for automatic determination of PU stage and PU texture based on its color.