Debridement of the surgical site during open fracture reduction and internal fixation is important for preventing surgical site infection; the risk of subsequent fracture-associated infection for a particular area of tissue is assessed by the surgeon based on multi-level variables, including demographics and laboratory results. Intraoperative fluorescence imaging can contribute additional information at a more localized level. Here we present a fluorescence-based predictive model using features from dynamic contrast enhanced-fluorescence imaging (DCE-FI), as well as patient-level variables associated with infection risk. Regions-of-interest were selected from thirty-eight enrolled open fracture patients. Spatial and kinetic features were extracted from DCE-FI, and combined with patient infection risk factor describing the possibility of getting surgical-site-infection. The model was evaluated for ability to predict composite outcome scores—intra-operative surgeon assessment coupled with post-operative confirmed infection outcome. This proposed model demonstrates high predictive performance with an accuracy of 0.86, evaluated with a cross-validation approach, and is a promising approach for early and quick identification of tissue prone to infection.
|