Deep Neural Network Classification of In Vivo Burn injuries with Different Etiologies Using THz Time-Domain Spectral Imaging

Patients arrive in burn centers with many etiologies of burn incidence (such as scald, contact, flame, steam, etc.). Treatment of these injuries depends on the burn depth and surface area of these burns. Moreover, earlier diagnosis and treatment of these burns lead to better patient outcomes. In this study, we show that THz-TDS imaging with a neural network classification algorithm can reproducibly classify burns, independent of their etiology, with an ROC-AUC of 88% in a porcine model.