Multimodal resting-state connectivity predicts affective neurofeedback performance

Abstract

Neurofeedback has been suggested as a potential complementary therapy to different psychiatric disorders. Of interest for this approach is the prediction of individual performance and outcomes. In this study, we applied functional connectivity-based modeling using electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) modalities to (i) investigate whether resting-state connectivity predicts performance during an affective neurofeedback task and (ii) evaluate the extent to which predictive connectivity profiles are correlated across EEG and fNIRS techniques. The fNIRS oxyhemoglobin and deoxyhemoglobin concentrations and the EEG beta and gamma bands modulated by the alpha frequency band (beta-m-alpha and gamma-m-alpha, respectively) recorded over the frontal cortex of healthy subjects were used to estimate functional connectivity from each neuroimaging modality. For each connectivity matrix, relevant edges were selected in a leave-one-subject-out procedure, summed into "connectivity summary scores" (CSS), and submitted as inputs to a support vector regressor (SVR). Then, the performance of the left-out-subject was predicted using the trained SVR model. Linear relationships between the CSS across both modalities were evaluated using Pearson's correlation. The predictive model showed a mean absolute error smaller than 20%, and the fNIRS oxyhemoglobin CSS was significantly correlated with the EEG gamma-m-alpha CSS (r = -0.456, p = 0.030). These results support that pre-task electrophysiological and hemodynamic resting-state connectivity are potential predictors of neurofeedback performance and are meaningfully coupled. This investigation motivates the use of joint EEG-fNIRS connectivity as outcome predictors, as well as a tool for functional connectivity coupling investigation.

Keywords: brain connectivity; electroencephalography; functional near-infrared spectroscopy; neurofeedback; resting-state.

FIGURE 1

(A) Schematic representation of EEG and fNIRS channels, and (B) an example of the cap on one volunteer. Anatomical representation of the positioning of (C) EEG electrodes, and (D) fNIRS optodes.

FIGURE 2

Schematic representation of the connectivity-based predictive modeling (adapted from Shen et al., 2017), summarized as: (A) Each participant has one connectivity matrix per neuroimaging feature and one performance measurement related to the affective neurofeedback task; (B) each edge in connectivity matrices is related to the subject’s performance using Pearson’s correlation in a leave-one-subject-out (LOSO) setup; (C) next, significant edges (p ≤ 0.05) are selected to create a mask; (D) then, for each subject in the training group, the mask is applied to select most important edges, which are then summarized (summed) into a single CSS value per subject; (E) next, an inter-subject vector is created for each neuroimaging modality, containing the respective summary values; (F) finally, these four vectors are used to train a support vector regressor (SVR) and test it in the subject-out.

FIGURE 3

Scatter plots illustrating the distribution of participants (dots) according to predicted performance (x-axis) and observed performance (y-axis). Red line represents the trend.

FIGURE 4

Averaged connectivity matrices showing the strength of the selected edges, and ring graphs describing the binary connections resulting from these edges for (A) oxyhemoglobin, (B) deoxyhemoglobin, (C) beta-m-alpha, (D) gamma-m-alpha. For connectivity matrices, hotter colors represent strong connectivity, while white squares represent connections removed by the correlation-based mask. For ring graphs, each black line represents a binary connection between two areas.

FIGURE 5

Scatter plots illustrating participants (dots) distribution according to (A) CSS from fNIRS oxyhemoglobin and deoxyhemoglobin, (B) CSS from EEG beta -m-alpha and gamma-m-alpha, (C) CSS from fNIRS oxyhemoglobin and EEG beta-m-alpha, (D) CSS from fNIRS oxyhemoglobin and EEG gamma-m-alpha, (E) CSS from fNIRS deoxyhemoglobin and EEG beta-m-alpha, and (F) CSS from fNIRS deoxyhemoglobin and EEG gamma-m-alpha. Red line represents the trend.