Reconstructing functional near-infrared spectroscopy (fNIRS) signals impaired by extra-cranial confounds: An easy-to-use filter method

- We conducted a simultaneous fNIRS-fMRI measurement with a working memory paradigm.
- In contrast to fMRI, fNIRS was not capable to find significant neural activation.
- Using extra-cranial fMRI we found fNIRS to be disturbed by extra-cranial signals.
- We propose a filter method to correct for disturbing, extra-cranial fNIRS signals.
- Filtered fNIRS showed significant activation in the expected brain region.

Functional near-infrared spectroscopy (fNIRS) is an optical neuroimaging method that detects temporal concentration changes of oxygenated and deoxygenated hemoglobin within the cortex, so that neural activation can be inferred. However, even though fNIRS is a very practical and well-tolerated method with several advantages particularly in methodically challenging measurement situations (e.g., during tasks involving movement or open speech), it has been shown to be confounded by systemic compounds of non-cerebral, extra-cranial origin (e.g. changes in blood pressure, heart rate). Especially event-related signal patterns induced by dilation or constriction of superficial forehead and temple veins impair the detection of frontal brain activation elicited by cognitive tasks. To further investigate this phenomenon, we conducted a simultaneous fNIRS-fMRI study applying a working memory paradigm (n-back). Extra-cranial signals were obtained by extracting the BOLD signal from fMRI voxels within the skin. To develop a filter method that corrects for extra-cranial skin blood flow, particularly intended for fNIRS data sets recorded by widely used continuous wave systems with fixed optode distances, we identified channels over the forehead with probable major extra-cranial signal contributions. The averaged signal from these channels was then subtracted from all fNIRS channels of the probe set. Additionally, the data were corrected for motion and non-evoked systemic artifacts. Applying these filters, we can show that measuring brain activation in frontal brain areas with fNIRS was substantially improved. The resulting signal resembled the fMRI parameters more closely than before the correction. Future fNIRS studies measuring functional brain activation in the forehead region need to consider the use of different filter options to correct for interfering extra-cranial signals.