Hyperspectral Imaging With Burn Contour Extraction for Burn Wound Depth Assessment

Abstract

Skin thermal burn wounds are classified according to subjective assessments of wound depth that indicate divergent modes of medical intervention. However, clinically discriminating superficial partial from deep partial thickness burns remains a significant challenge, where only the latter requires excision and skin grafting. Motivated by the need for and ramifications of an objective burn wound assessment tool, this paper advances hyperspectral imaging (HSI) in a porcine skin burn model to quantitatively evaluate thermal burn injuries (superficial and deep partial thickness burns). Two-dimensional (2D) principal component analysis for noise reduction is applied to images captured by HSI in the visible wavelength range. Herein, a multivariate regression analysis is used to calculate the total hemoglobin concentration (tHb) and the oxygen saturation (StO2) of the injured tissue. These perfusion profiles are spatially mapped to yield characteristic distributions corresponding to the burn wound degree validated histologically. The results demonstrate that StO2 and tHb diverge significantly for superficial partial and deep partial burns at 24 h and 1 h, respectively (p < 0.05). A StO2 burn map at 1 h post-burn yields a 2D burn contour that is registered with a burn color image. This early stage burn-specific contour has implications to guide downstream burn excision and grafting.