Two-detector Corrected Near Infrared Spectroscopy (C-NIRS) detects hemodynamic activation responses more robustly...

In near-infrared spectroscopy (NIRS) of human cerebral hemodynamics, detection of stimulus-related responses is confounded by the presence of unrelated trends in both the brain and the overlying scalp. A proposed strategy for reducing hemodynamic noise has been to record "scalp only" trends simultaneously via a second shorter-separation detector (~5 mm rather than ~30 mm) and perform a subtraction (C-NIRS, for "corrected near-infrared spectroscopy"). To compare the single- and dual-detector strategies, a 21-volunteer study of visual stimulation responses (6 stimulation blocks and 8 recording channels per measurement run) has been conducted. Activation-flagged channels were defined based upon (a) the significance (p-value) of the average rise in oxyhemoglobin concentration and (b) the average signal-to-noise over 6 stimulation epochs. At reasonable thresholds (p<0.025, SNR>1), the C-NIRS method increased the number of activation-flagged channels from 47 to 66, an increase of 40%, adding 24 channels and eliminating only 5. Of the 71 channels that were activation-flagged by at least one modality, the C-NIRS time series exhibited more significant oxyhemoglobin rise in 80% of such channels, and better signal-to-noise in 73%. In addition, single-subject C-NIRS stimulus responses were more consistent than NIRS over the six stimulation epochs, with significantly lower coefficients of variation in both amplitude and latency (i.e. time between stimulus onset and maximum hemoglobin rise). These results demonstrate that two-detector C-NIRS provides a straightforward way of (a) removing hemodynamic interference from NIRS data, (b) increasing the detection rate of cerebrally-unique responses, and (c) improving the quality of those recorded responses. Parallel insights regarding deoxyhemoglobin trends could not be drawn from this data set but should be attainable in future studies with higher signal to noise ratios.