Wearable functional near-infrared spectroscopy (fNIRS) permits cortical hemodynamic imaging in freely moving subjects, but the oxygenated-hemoglobin (HbO2) signal it relies on is dominated by extraneous systemic physiological noise — blood-pressure fluctuations and Mayer waves, respiration, and cardiac pulsation — carried in the superficial layers of the head. The study that first demonstrated wearable multi-channel fNIRS during an outdoor activity (Piper et al., 2014) identified as its central unresolved issue that no common standard existed for dealing with this biological noise, leaving candidate approaches (multi-distance nearest-neighbor regression, principal component analysis, general linear model methods, adaptive filtering) unevaluated. A decade later, short-separation regression has become the leading candidate and community best-practice guidelines recommend it, yet the most recent empirical work (Lemaire et al., 2026) still finds that methodological standardization remains a challenge, with the major analysis toolboxes implementing short-channel correction differently. The open problem is to establish and validate a common standard for systemic-noise removal in mobile/wearable fNIRS — a method or explicit condition-dependent rule demonstrated to remove systemic interference from HbO2 (and HbR) without distorting the evoked cortical response, across freely moving paradigms, populations, and hardware.

Background

Functional near-infrared spectroscopy (fNIRS) is a noninvasive optical neuroimaging technique that estimates changes in cortical concentrations of oxygenated and deoxygenated hemoglobin (HbO2 and HbR) by illuminating the head with near-infrared light (roughly 700–900 nm, where hemoglobin absorbs) and detecting the light that scatters back to sensors on the scalp. Because the light must traverse scalp and skull, the measured signal always mixes the cortical hemodynamic response with systemic physiological fluctuations of non-cerebral origin: cardiac pulsation (around 1 Hz), respiration (around 0.2–0.3 Hz), slow blood-pressure oscillations known as Mayer waves (around 0.1 Hz), and slower blood-pressure drifts. These extraneous biological signals travel mainly in the superficial tissue layers and disproportionately corrupt the HbO2 signal, because HbO2 fluctuations in scalp vasculature are large compared with the small evoked cortical response. Most fNIRS systems around 2010 additionally tethered the subject to a benchtop instrument with optical fiber bundles, confining experiments to the laboratory. Piper et al. (2014) presented a wearable, fiberless, multi-channel fNIRS system (eight dual-wavelength light-emitting diodes and eight electro-optical sensors, about 25 mm source–detector spacing) and demonstrated motor-cortex imaging during left-hand gripping in N=8 subjects under three conditions — outdoor bicycle riding, stationary pedaling, and sitting still — the first functional NIRS brain imaging during an outdoor activity in a real-life situation. A significant task-related HbR decrease over the contralateral motor cortex appeared in all three conditions, while a significant task-related HbO2 increase survived only when the subject was not pedaling, illustrating how strongly systemic physiology and movement confound the HbO2 signal in mobile recordings. In their Discussion the authors note that instrumental and signal-analysis schemes exist to reduce the interference from global hemodynamics — multi-distance measurements for nearest-neighbor regression, principal component analysis (PCA, which removes the dominant variance components on the assumption they are noise), general linear model (GLM) methods (fitting the time series as a linear combination of an expected-response design matrix plus nuisance regressors), and adaptive filtering (subtracting a noise reference weighted by continuously updated coefficients) — but state that no common standard had yet been established in dealing with extraneous biological noise and that evaluating these approaches was beyond the scope of their manuscript. Multi-distance regression has since matured into short-separation regression: channels with source–detector separations of roughly 1–1.5 cm sample only superficial tissue, and can be regressed out of the longer channels that reach cortex (Gagnon et al. 2012, 2014). Other limitations remarked on elsewhere in the paper (the rather limited number of subjects and trials, empirically chosen channel/trial rejection thresholds, penetration depth and optode-spacing-limited spatial resolution, contact-based interference and tethering, more complex or dual tasks and brain regions beyond sensorimotor cortex, and combination with EEG) are device-validation limitations of this specific 2014 prototype that later mobile-fNIRS work has pursued and largely overtaken; they are not part of the methodological open problem retained here.

Problem Statement

Piper et al. (2014) state that no common standard has been established for dealing with extraneous biological noise — systemic blood pressure, respiration, cardiac pulsation, Mayer waves — that primarily corrupts the HbO2 signal in wearable/mobile fNIRS, and that evaluating the candidate approaches (multi-distance nearest-neighbor regression, principal component analysis, general linear model methods, adaptive filtering) was beyond the scope of their study. The problem is to settle this question: determine, by systematic and reproducible comparison, which noise-rejection strategy — a single method, a combination, or an explicit condition-dependent decision rule — most effectively removes systemic physiological interference from HbO2 (and HbR) measurements in wearable/mobile fNIRS without distorting the evoked cortical hemodynamic response, and validate that strategy in freely moving paradigms (such as locomotion or naturalistic behavior) so that it can serve as a common standard for the field.

References

  1. Piper SK, Krueger A, Koch SP, Mehnert J, Habermehl C, Steinbrink J, Obrig H, Schmitz CH. A wearable multi-channel fNIRS system for brain imaging in freely moving subjects. NeuroImage 85 Pt 1:64–71 (2014). DOI: 10.1016/j.neuroimage.2013.06.062. https://doi.org/10.1016/j.neuroimage.2013.06.062 (open-access author manuscript: PMCID PMC3859838)
  2. Gagnon L, Cooper RJ, Yücel MA, Perdue KL, Greve DN, Boas DA. Short separation channel location impacts the performance of short channel regression in NIRS. NeuroImage 59:2518–2528 (2012). DOI: 10.1016/j.neuroimage.2011.08.095. https://doi.org/10.1016/j.neuroimage.2011.08.095
  3. Gagnon L, Yücel MA, Boas DA, Cooper RJ. Further improvement in reducing superficial contamination in NIRS using double short separation measurements. NeuroImage 85:127–135 (2014). DOI: 10.1016/j.neuroimage.2013.01.073. https://doi.org/10.1016/j.neuroimage.2013.01.073
  4. Brigadoi S, Ceccherini L, Cutini S, Scarpa F, Scatturin P, Selb J, Gagnon L, Boas DA, Cooper RJ. Motion artifacts in functional near-infrared spectroscopy: A comparison of motion correction techniques applied to real cognitive data. NeuroImage 85:181–191 (2014). DOI: 10.1016/j.neuroimage.2013.04.082. https://doi.org/10.1016/j.neuroimage.2013.04.082
  5. Fishburn FA, Ludlum RS, Vaidya CJ, Medvedev AV. Temporal Derivative Distribution Repair (TDDR): A motion correction method for fNIRS. NeuroImage 184:171–179 (2019). DOI: 10.1016/j.neuroimage.2018.09.025. https://doi.org/10.1016/j.neuroimage.2018.09.025
  6. Pinti P, Scholkmann F, Hamilton A, Burgess P, Tachtsidis I. Current Status and Issues Regarding Pre-processing of fNIRS Neuroimaging Data: An Investigation of Diverse Signal Filtering Methods Within a General Linear Model Framework. Frontiers in Human Neuroscience 12:505 (2019). DOI: 10.3389/fnhum.2018.00505. https://doi.org/10.3389/fnhum.2018.00505
  7. Yücel MA, von Lühmann A, Scholkmann F, Gervain J, Dan I, Ayaz H, Boas D, Cooper RJ, Culver J, Elwell CE, Eggebrecht A, Franceschini MA, Grova C, Homae F, Lesage F, Obrig H, Tachtsidis I, Tak S, Tong Y, Torricelli A, Wabnitz H, Wolf M. Best practices for fNIRS publications. Neurophotonics 8(1):012101 (2021). DOI: 10.1117/1.nph.8.1.012101. https://doi.org/10.1117/1.nph.8.1.012101
  8. Menant JC, Maidan I, Alcock L, Al-Yahya E, Cerasa A, Clark DJ, de Bruin ED, Fraser S, Gramigna V, Hamacher D, Herold F, Holtzer R, Izzetoglu M, Lim S, Pantall A, Pelicioni P, Peters S, Rosso AL, St George R, Stuart S, Vasta R, Vitorio R, Mirelman A. A consensus guide to using functional near-infrared spectroscopy in posture and gait research. Gait & Posture 82:254–265 (2020). DOI: 10.1016/j.gaitpost.2020.09.012. https://doi.org/10.1016/j.gaitpost.2020.09.012
  9. Pinti P, Tachtsidis I, Hamilton A, Hirsch J, Aichelburg C, Gilbert S, Burgess PW. The present and future use of functional near-infrared spectroscopy (fNIRS) for cognitive neuroscience. Annals of the New York Academy of Sciences 1464(1):5–29 (2020). DOI: 10.1111/nyas.13948. https://doi.org/10.1111/nyas.13948
  10. Pinti P, Dina LM, Smith TJ. Ecological functional near-infrared spectroscopy in mobile children: using short separation channels to correct for systemic contamination during naturalistic neuroimaging. Neurophotonics 11(4):045004 (2024). DOI: 10.1117/1.nph.11.4.045004. https://doi.org/10.1117/1.nph.11.4.045004
  11. Lemaire Y, Barone P, Strelnikov K, Deguine O, Caclin A, Ginzburg J. Optimizing short-channel regression in fNIRS: an empirical evaluation with ecological audiovisual stimuli. Neurophotonics 13(1):015012 (2026). DOI: 10.1117/1.nph.13.1.015012. https://doi.org/10.1117/1.nph.13.1.015012