Optimizing the general linear model for functional near-infrared spectroscopy: an adaptive hemodynamic response function approach

Abstract

Abstract. An increasing number of functional near-infrared spectroscopy (fNIRS) studies utilize a general linear model (GLM) approach, which serves as a standard statistical method for functional magnetic resonance imaging (fMRI) data analysis. While fMRI solely measures the blood oxygen level dependent (BOLD) signal, fNIRS measures the changes of oxy-hemoglobin (oxy-Hb) and deoxy-hemoglobin (deoxy-Hb) signals at a temporal resolution severalfold higher. This suggests the necessity of adjusting the temporal parameters of a GLM for fNIRS signals. Thus, we devised a GLM-based method utilizing an adaptive hemodynamic response function (HRF). We sought the optimum temporal parameters to best explain the observed time series data during verbal fluency and naming tasks. The peak delay of the HRF was systematically changed to achieve the best-fit model for the observed oxy-and deoxy-Hb time series data. The optimized peak delay showed different values for each Hb signal and task. When the optimized peak delays were adopted, the deoxy-Hb data yielded comparable activations with similar statistical power and spatial patterns to oxy-Hb data. The adaptive HRF method could suitably explain the behaviors of both Hb parameters during tasks with the different cognitive loads during a time course, and thus would serve as an objective method to fully utilize the temporal structures of all fNIRS data.

Comparing diffuse optical tomography and functional magnetic resonance imaging signals during a cognitive task: pilot study

“…Polarities of the canonical HRF functions were reversed to visualize negative response. 30 After estimation, activation maps were generated for the subjects by applying a fixed effects model analysis. Three contrasts were computed: difficult count × easy count, difficult count > rest, and easy count > rest.…”

Real-Time Reduction of Task-Related Scalp-Hemodynamics Artifact in Functional Near-Infrared Spectroscopy with Sliding-Window Analysis

“…Our proposed method, using a semi-real simulation with the assumption that fNIRS signals were possibly contaminated with global task-related SBF, achieved a significantly higher estimation accuracy than conventional design matrices of GLM, which have been used in several studies [5,16,[26][27][28][29]. However, brain activity estimations with real fNIRS data are more challenging.…”

“…and standard (CHM + CHM (1) + CHM (2) + Const.) [5,16,[26][27][28][29]. Both of the comparative design matrices were also applied to SWA for real-time CBF estimation.…”

Neuroimaging with functional near infrared spectroscopy: From formation to interpretation

“…For other neuroimaging modalities it may represent an idealization of the action potential. For fNIRS it takes the form of an hemodynamic response function [35,161,139,137] that is sometimes convolved with the neural kernel [144]. This hemodynamic response function is known to be non-linear [162] and varies with age [163].…”

“…Analysis approaches can be coarsely split into hypothesis-driven or data-driven depending on whether the hypothesis (model) is imposed a priori or a posteriori. Common operations to decode activity include regressions, most times in the form of the general linear model (GLM) [138,139], correlations [122], frequency coherence [140] and wavelet coherence [137], statistical inference [141] among others. The GLM is described in Eq 8, where y θ k,x ∈ R N are temporal samples, θ k,x ∈ R N denotes the noise, X ∈ R N xM is the design matrix and β θ k,x ∈ R M is the response signal strength.…”