Table 2.

Recommendations for fNIRS application, fNIRS data processing and fNIRS data analysis

fNIRS recording
Optode placement	Optimal solution: ▪Use a neuronavigational approach Alternative solution: ▪ Use 10-20 (10-10 or 10-5) international EEG-system ➢ If MRI scan is possible → Co-registration ➢ If MRI scan is not possible → Registration via 3-D-Digitizer or                → Virtual spatial (probabilistic) registration
Source–detector separation	▪At least 3.0 cm for “long-separation channels” ▪Around 0.8 cm for “short-separation channels”
Baseline recording	▪Record baseline in sitting position ▪Choose an appropriate baseline duration (e.g., with regard to study design) ▪Ensure that the fNIRS channels have a good SNR (e.g., look for blood volume pulsation)
fNIRS data processing: conversion and artefact removal
Conversion of optical density changes into concentration changes of chromophores (e.g. oxyHb, deoxyHb, totHb)	▪Apply modified Beer–Lambert law with appropriate $\mu \mathrm{a}$ and DPF values
- DPF value determination	Optimal solution: ▪Direct quantification of DPF values using frequency- or time-domain fNIRS Alternative solution: ▪Use formulas allowing the calculation of individual, age-specific, and wavelength-specific DPF values
Artefact removal
Removal of motion artefacts *	▪Use of high-performing methods (e.g., Wavelet filtering or hybrid filter methods)
Removal of physiological artefacts	▪Use of high-performing methods (e.g., SDS regression to filter out extracerebral signal components)
General artefact removal	▪Use a band-pass filtering with appropriate cut-off frequencies (e.g. considering stimulus or task paradigm)
fNIRS data processing: further analysis
Detrending	▪Perform baseline correction or normalization
Analysis	▪Perform averaging across channels and trials or perform GLM analysis # ▪Choose an appropriate temporal window (e.g., consider delay in hemodynamic responses) ▪Use at least oxyHb and deoxyHb for statistical analysis

deoxyHb: deoxygenated hemoglobin; DPF: differential path length factor; EEG: electroencephalography; fNIRS: functional near-infrared spectroscopy; GLM: general linear model; $\mu \mathrm{a}$: absorption coefficient; MRI: magnetic resonance imaging; oxyHb: oxygenated hemoglobin; SDS: short-separation channel (also known as short-distance channel); SNR: signal-to-noise ratio/* Filtering of motion artefacts can also be conducted on optical density data (before conversion into concentration changes) depending on the used filter methods and/or software solution. / ^# Please note, if distinct types of GLM are used (e.g., GLM with model correction methods) the processing steps are divergent from those shown in the table and some of the given recommendations do not apply in this particular case.