Maximizing the Reliability and Precision of Measures of Prefrontal Cortical Oxygenation Using Frequency-Domain Near-Infrared Spectroscopy

Abstract

Frequency-domain near-infrared spectroscopy (FD-NIRS) has been used for non-invasive assessment of cortical oxygenation since the late 1990s. However, there is limited research demonstrating clinical validity and general reproducibility. To address this limitation, recording duration for adequate validity and within- and between-day reproducibility of prefrontal cortical oxygenation was evaluated. To assess validity, a reverse analysis of 10-min-long measurements (n = 52) at different recording durations (1–10-min) was quantified via coefficients of variation and Bland–Altman plots. To assess within- and between-day within-subject reproducibility, participants (n = 15) completed 2-min measurements twice a day (morning/afternoon) for five consecutive days. While 1-min recordings demonstrated sufficient validity for the assessment of oxygen saturation (StO2) and total hemoglobin concentration (THb), recordings ≥4 min revealed greater clinical utility for oxy- (HbO) and deoxyhemoglobin (HHb) concentration. Females had lower StO2, THb, HbO, and HHb values than males, but variability was approximately equal between sexes. Intraclass correlation coefficients ranged from 0.50–0.96. The minimal detectable change for StO2 was 1.15% (95% CI: 0.336–1.96%) and 3.12 µM for THb (95% CI: 0.915–5.33 µM) for females and 2.75% (95%CI: 0.807–4.70%) for StO2 and 5.51 µM (95%CI: 1.62–9.42 µM) for THb in males. Overall, FD-NIRS demonstrated good levels of between-day reliability. These findings support the application of FD-NIRS in field-based settings and indicate a recording duration of 1 min allows for valid measures; however, data recordings of ≥4 min are recommended when feasible.