Semiautomatic classification of secondary healing ulcers in multispectral images

Close-up color photographs of the wounds were used as imput. The interesting areas were marked by the operator, creating a binary image. According to the severity of the wound, one of the trained classifiers was selected. The digital picture was classified and combined with the binary image, giving the qualitative (proportion necroses/fibrin and granulation) and the quantitative (wound size) parameters. In this way the time for analyzing a large number of wound photographs was substantially reduced. The method has been programmed in Pascal on the IMTEC Epsilon image processing system