The purpose of the present study was to assess the retest reliability of cortical activation detected by event-related functional near-infrared spectroscopy (fNIRS) based on craniocerebral correlations. Isolated functional activation was evoked in the motor cortex by a periodically performed finger-tapping task. During 44-channel fNIRS recording, 12 subjects performed 30 trials of right and left index finger tapping in two sessions. The retest interval was set to 3 weeks. Simple correlations of the contrast t-values supplemented by scatterplots, channel-wise intraclass correlation coefficients (ICC), as well as reproducibility indices for the size and the location of the detected activation were calculated. The results at the group level showed sufficient single measure ICCs (up to 0.80) and excellent reproducibility of the size and the location (up to 89% were reproducible). Comparisons of the intersession group amplitudes demonstrate that the fNIRS signals were stable across time in a retest study design: the number of significant differences was less than randomly occurring false-positive activated channels if an alpha level of 5% is chosen. Effect size analyses indicated that the intersession amplitude differences are small (mean < 0.25). For deoxyhemoglobin and oxyhemoglobin distinct statistical power profiles were revealed regarding the activation vs. baseline contrast as well as the intersession amplitude differences, indicating a higher sensitivity of deoxyhemoglobin for local hemodynamic changes. The results suggest that sensorimotor activation assessed by event-related fNIRS based on craniocerebral correlations is sufficiently reproducible at the group level.
2006 Wiley-Liss, Inc