Written and maintained by CASRAI Editorial Board
Last updated
Short Answer
Functional Near-Infrared Spectroscopy (fNIRS) is a non-invasive optical neuroimaging modality that measures cortical hemodynamic activity in the human brain. By projecting near-infrared light through the scalp and skull, fNIRS detects concentration changes in oxygenated and deoxygenated hemoglobin driven by neurovascular coupling, providing a portable, motion-tolerant alternative to functional magnetic resonance imaging.
Definition and Scope
Functional near-infrared spectroscopy evaluates localized cerebral blood oxygenation using low-energy optical radiation. The technique capitalizes on the biological optical window in the near-infrared spectrum, spanning approximately 650 to 900 nanometers. Within this spectral band, biological tissues—including skin, cranial bone, and cerebrospinal fluid—exhibit minimal optical absorption, allowing photons to diffuse several centimeters into tissue.
While tissue scattering is pronounced, the primary endogenous absorbers (chromophores) in the head are oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR). When localized neuronal populations fire in response to cognitive or sensory tasks, localized blood flow surges to supply glucose and oxygen (functional hyperemia). This hemodynamic response causes a transient rise in local HbO accompanied by a decrease in HbR. Within academic neuroscience, fNIRS bridges the gap between electroencephalography (EEG) and functional MRI, delivering continuous hemodynamic tracking in naturalistic environments.
How It Works and How It Is Used in Practice
Implementing an fNIRS experiment involves optode positioning, optical signal conversion, and physiological noise filtering.
The Modified Beer-Lambert Law (MBLL)
In non-scattering media, spectrophotometry quantifies chemical concentration through the Beer-Lambert law. In biological tissue, however, multiple elastic scattering causes photons to travel along convoluted, banana-shaped trajectories between light sources and detectors. To account for this lengthened path, fNIRS uses the Modified Beer-Lambert Law:
ΔOD(λ) = -ln(I(λ) / I₀(λ)) = [ε_HbO(λ) · Δ[HbO] + ε_HbR(λ) · Δ[HbR]] · d · DPF(λ)
Here, ΔOD is optical density change at wavelength λ, I₀ is baseline light intensity, I is detected intensity, ε is the molar extinction coefficient, d is geometrical source-detector distance (standardly ~30 mm), and DPF is the unitless Differential Pathlength Factor. Because HbO and HbR have distinct absorption spectra crossing at the isosbestic point (~805 nm), measuring attenuation at two or more paired wavelengths (e.g., 760 nm and 850 nm) allows the matrix equation to be solved, yielding micromolar changes in [HbO] and [HbR].
The Research Workflow and Preprocessing Pipeline
A rigorous fNIRS analysis pipeline, endorsed by the Society for Functional Near-Infrared Spectroscopy (SfNIRS) and implemented in MNE-Python and Homer3, follows standard analytical steps:
- Optode Array Placement: Emitter and detector optodes are secured in an elastic cap aligned to the international 10–20 system. Source-detector separations are fixed at 25–35 mm to achieve sensitivity to the outer cerebral cortex (~15 mm depth).
- Raw Intensity to Optical Density: Continuous optical raw intensity measurements are converted to relative optical density changes.
- Motion Artifact Correction: Head movements induce transient optode decoupling, appearing as high-frequency baseline shifts or spikes. Algorithms like spline interpolation, wavelet filtering, or targeted PCA correct these mechanical artifacts without eliminating biological signals.
- Superficial Physiological Denoising: Optical channels capture both cortical hemodynamic responses and systemic vascular fluctuations in the scalp (cardiac pulsations ~1 Hz, respiration ~0.25 Hz, Mayer waves ~0.1 Hz). The gold standard for isolation is incorporating dedicated short-separation channels (~8 mm separation). Because short channels penetrate only the scalp, their signal is regressed out of the standard 30 mm cortical channels via General Linear Model (GLM) filtering.
- Hemodynamic Concentration Conversion: The Modified Beer-Lambert Law converts filtered optical density into relative micromolar concentrations of HbO and HbR.
- Statistical Inference: Channel-wise GLM estimation tests task-related activation contrasts, controlling for multiple comparisons using False Discovery Rate (FDR) adjustments. Standardized data can be curated in the open SNIRF (Shared Near Infrared Spectroscopy Format) format and archived in community databases like OpenNeuro.
Types and Instrumentation Architectures
Functional near-infrared spectroscopy systems are classified into three distinct architectures based on light emission and detection technology:
| System Architecture | Light Source Technology | Measured Quantities | Key Advantages | Operational Limits |
|---|---|---|---|---|
| Continuous Wave (CW-fNIRS) | LEDs or CW laser diodes | Light intensity attenuation (amplitude) | Low cost, highly portable, compact wearable caps, battery-powered | Cannot measure absolute pathlength; assumes constant DPF; relative metrics only |
| Frequency Domain (FD-fNIRS) | Intensity-modulated lasers (tens to hundreds of MHz) | Amplitude attenuation and phase shift | Separates absorption from scattering coefficients; absolute baseline quantification | Higher hardware expense and complexity; bulkier optode cables |
| Time-Resolved (TR-fNIRS) | Pulsed lasers and single-photon detectors | Temporal point spread function (photon arrival time) | Direct depth discrimination; separates superficial scalp from cortical signals directly | Highest hardware cost; complex calibration; bulky benchtop footprint |
Where It Matters for Research and Laboratories
Because fNIRS is lightweight, silent, and resilient to participant movement, it enables experimental paradigms that are unfeasible in enclosed MRI scanners:
- Developmental Cognitive Neuroscience: Infant brain development is widely studied with fNIRS because infants can sit naturally on a caregiver’s lap while looking at stimuli, avoiding sedation or physical restraint required by MRI.
- Naturalistic Movement and Ergonomics: Wearable, battery-operated fNIRS headsets record cortical activation during outdoor walking, manual industrial tasks, flight simulation, and athletic training.
- Social Hyperscanning: Multiple fNIRS headsets operating synchronously allow researchers to simultaneously record brain activity from two or more communicating partners during live conversation or cooperative negotiation.
- Translational Medicine and Neurorehabilitation: Tracking longitudinal recovery of motor and prefrontal function in stroke and traumatic brain injury during active physical therapy, advancing empirical links across neuroscience and neurobiology.
Common Misunderstandings
- fNIRS Does Not Measure Electrical Activity Directly: Unlike EEG, which detects millivolt electrical potentials produced directly by postsynaptic neuronal currents, fNIRS measures the downstream hemodynamic response (seconds-scale metabolic blood flow changes) via neurovascular coupling.
- fNIRS Cannot Image Deep Subcortical Structures: Because photon diffusion depth is physically constrained to roughly half the source-detector distance (approximately 1.2 to 1.5 cm into the head), fNIRS records only from the superficial cerebral cortex. Subcortical structures such as the hippocampus, amygdala, and thalamus remain undetectable.
- Scalp Blood Flow Is Not Brain Activity: Without short-separation reference channels or systemic filtering, changes in heart rate, blood pressure, or forehead sweating during cognitive stress can masquerade as cortical activation, leading to erroneous research conclusions.
Frequently Asked Questions
What is the spatial resolution of fNIRS?
The spatial resolution of fNIRS is typically 1 to 2 centimeters for standard continuous-wave channel layouts. While lower than functional MRI (which achieves 1 to 3 millimeters), spatial resolution can be improved to approximately 5 to 10 millimeters using high-density diffuse optical tomography (HD-DOT) arrays with overlapping multi-distance channels.
How much does an fNIRS system cost?
Research-grade fNIRS systems range in cost from approximately $15,000 to $40,000 for portable 8- to 16-channel headsets, and from $80,000 to $250,000 or more for modular, high-density whole-head systems with 64 to 128 channels. In contrast, acquiring and maintaining a research 3T MRI scanner requires millions of dollars in capital expenditure and dedicated shielding infrastructure.
What are the limitations of fNIRS?
The primary limitations of fNIRS include restricted depth penetration (limited to the outer 1.5 cm of cortical grey matter), moderate spatial resolution (~1–2 cm), optical signal attenuation caused by dense or darkly pigmented hair, and vulnerability to superficial extracerebral vascular contamination in the scalp.
What is the newest brain scanning technique?
Recent innovations in optical neuroimaging include High-Density Diffuse Optical Tomography (HD-DOT)—which uses hundreds of densely overlapping channels to reconstruct volumetric images approaching fMRI spatial resolution—as well as wireless fiberless wearable fNIRS caps and hybrid fNIRS-EEG systems that combine hemodynamic and electrophysiological tracking simultaneously.
What are the uses of near-infrared spectroscopy?
Near-infrared spectroscopy is utilized across academic neuroscience for cognitive tracking, infant developmental studies, naturalistic mobile ergonomics, and social hyperscanning. In clinical hospital settings, continuous NIRS sensors monitor regional cerebral tissue oxygenation (cerebral oximetry) during neonatal intensive care, cardiovascular bypass surgery, and neurological trauma management.
Sources
- Yücel, M. A., Lühmann, A. V., Scholkmann, F., Gervain, J., Dan, I., et al. (2021). Best practices for fNIRS publications. Neurophotonics, 8(1), 012101. https://doi.org/10.1117/1.NPh.8.1.012101 (read via PubMed efetch 2026-11-08).
- Scholkmann, F., Kleiser, S., Metz, A. J., Zimmermann, R., Mata Pavia, J., Wolf, U., & Wolf, M. (2014). A review on continuous wave functional near-infrared spectroscopy and imaging instrumentation and methodology. NeuroImage, 85 Pt 1, 6–27. https://doi.org/10.1016/j.neuroimage.2013.05.004 (read via PubMed efetch 2026-11-08).
- MNE-Python Consortium. Preprocessing functional near-infrared spectroscopy (fNIRS) data. https://mne.tools/stable/auto_tutorials/preprocessing/70_fnirs_processing.html (read 2026-11-08).
- Society for Functional Near-Infrared Spectroscopy. OpenfNIRS community standards and data specifications. https://openfnirs.org/ (read 2026-11-08).
