Cerebral hemoglobin and optical pathlength influence near-infrared spectroscopy measurement of cerebral oxygen saturation

Near-infrared spectroscopy (NIRS) is a noninvasive optical technique to monitor cerebral oxygen saturation at the bedside. Despite its applicability, NIRS has had limited clinical use because of concerns about accuracy, noted by intersubject variability in slope and intercept of the line between NIRS- and weighted-average arterial-cerebrovenous saturation (SMO2). This study evaluated transcranial optical pathlength and cerebral hemoglobin concentration as sources for this intersubject variability. Experiments were performed in an in vitro brain model and in piglets. Optical pathlength and cerebral hemoglobin concentration were measured by time-resolved spectroscopy (TRS). NIRS and TRS were recorded in the model, as perfusate blood saturation was varied (0%-100%) at several hemoglobin concentrations, and in piglets, as SMO2 was varied (15%-90%) before and after hemodilution. In the model, hemoglobin concentration significantly altered the NIRS versus blood saturation line slope and intercept, as well as optical pathlength. In piglets (before hemodilution), there was significant intersubject variability in NIRS versus SMO2 line slope (0.73-1.4) and intercept (-24 to 36) and in transcranial optical pathlength (13.4-16 cm) and cerebral hemoglobin concentration (0.58-1.1 g/dL). By adjusting the NIRS algorithm with optical pathlength or cerebral hemoglobin measurements, intersubject variability in slope (0.9-1.2) and intercept (-9 to 18) decreased significantly. Hemodilution significantly changed NIRS versus SMO2 line slope and intercept, as well as transcranial optical pathlength and cerebral hemoglobin concentration (before versus after hemodilution: slope 0.9 vs 0.78, intercept 13 vs 19, pathlength 13.9 vs 15.6 cm, cerebral hemoglobin 0.98 vs 0.73 g/dL). By adjusting the NIRS algorithm with the cerebral hemoglobin measurements, slope and intercept remained unchanged by hemodilution. These data indicate that intersubject variability in NIRS originates, in part, from biologic variations in transcranial optical pathlength and cerebral hemoglobin concentration. Instruments to account for these factors may improve NIRS cerebral oxygen saturation measurements.