Estimation of core body temperature by near-infrared imaging of vein diameter change in the dorsal hand

Core body temperature (T-core) is a key indicator of personal thermal comfort and serves as a monitor of thermal strain. Multi-parametric sensors are not practical for estimating core temperature because they require long data collection times and a wide variety of settings. This study introduces dorsal hand vein dynamics as a novel indicator along with heart rate (HR) and dorsal hand skin temperature (T-hand) for predicting T-core during rest following T-core elevation. Twelve healthy males aged 27 +/- 9 years old participated in the experiment. The experimental procedure consisted of a 10-min rest followed by 60 min of passive heat stress induced by leg immersion in hot water at 42 degrees C and a 40-min thermal relaxation period after the legs were removed from the water. A near-infrared (NIR) imaging system was configured to monitor the dorsal hand veins during the entire experimental session. The values of HR, T-hand, and T-core were continuously monitored while the ambient temperature and relative humidity (RH) were maintained in a climate chamber at 20 degrees C and 50%, respectively. Our selected predictor parameters demonstrated similar patterns in the T-core such that the value increased as a result of passive heat stress and decreased in the thermal relaxation phase. The experimental data were divided into two phases: thermal stress and relaxation. At the resting condition, inclusion of the hand vein diameter (VD) improved the multiple linear regression value (R-2) about 26%. At the relaxation phase, however, training regressions R-2 = 0.68 and R-2 = 0.94 were observed in the regression model with and without considering VD, respectively. The test regression value of R-2 = 0.88 and the root mean square error (RMSE) of 0.18 degrees C showed good agreement with the predicted values. These findings demonstrate acceptable validity of the non-invasive T-core estimation at the resting condition. In particular, the inclusion of VD as a predictor in the regression analysis increases the prediction accuracy with a lower RMSE value. (C) 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement