Functional near-infrared spectroscopy: a long-term reliable tool for measuring brain activity during verbal fluency.

The present study investigated the short- and long-term retest reliability of brain activity measured with functional near-infrared spectroscopy (fNIRS) during verbal fluency, the most published cognitive task within fNIRS literature. We examined 15 healthy right handed subjects in a block design task with retest intervals of three weeks and one year. Performance was constant over time. Amplitude of brain activation, as indicated by increases of oxygenated (O(2)Hb) and total (totHb) and decreases of deoxygenated haemoglobin (HHb), was reduced at session two and reversed at the third session for the fluency related region of interest (ROI). Small decreases for session two and three were found outside the ROI. These changes in amplitude may contribute to variability of reproducibility as measured with intraclass correlation coefficients (ICCs) within the ROI. Acceptable reliability for all chromophores and comparisons was reached for the mean of repeated measures at cluster level. Spatial (size and localisation), temporal and whole probe set activity was completely acceptable without exception. Retest reliability was not satisfactory at single subject and single channel level. Amplitude decreases over time outside the ROI suggest higher physiological or arousal effects for session one. Amplitude recovery in the ROI in session three argues for a psychological effect. Overall our findings indicate that fNIRS analyses at single subject and single channel level should be interpreted cautiously, while group and cluster analyses have sufficient test retest reliability.