CIELAB based system for burn depth estimation

摘要

Successful cure of a burn injury depends highly on the first treatment. Burn depth is traditionally defined in three degrees. Estimation of depth degree is carried out by visual evaluation of the wound by the specialized dermatological experts. This type of evaluation includes a high degree of subjectivity. In the literature it can found objective methods for determining the depth of the burn by processing of digital photographic images. However, these methods estimate only one degree per burn wound despite the fact that it is common to find all three types within the same burn wound. In this paper a classification system to estimate the different depth degrees that a burn wound can present, is proposed. A color characterization algorithm is initially applied to the photographic images. A color and texture features extraction based on the L*a*b* color space and the chromatic opponent channels is carried out. The classifier used is a Fuzzy-ARTMAP neural network. This neural network performs a pixel-based classification to estimate the different depth degrees present in burn wound image. The system has been tested with 60 images. A success rate of around 80% has been achieved.