Closed-Loop Infraslow Brain–Computer Interface can Modulate Cortical Activity and Connectivity in Individuals With Chronic Painful Knee Osteoarthritis: A Secondary Analysis of a Randomized Placebo-Controlled Clinical Trial

Abstract

Introduction. Chronic pain is a percept due to an imbalance in the activity between sensory-discriminative, motivational-affective, and descending pain-inhibitory brain regions. Evidence suggests that electroencephalography (EEG) infraslow fluctuation neurofeedback (ISF-NF) training can improve clinical outcomes. It is unknown whether such training can induce EEG activity and functional connectivity (FC) changes. A secondary data analysis of a feasibility clinical trial was conducted to determine whether EEG ISF-NF training can significantly alter EEG activity and FC between the targeted cortical regions in people with chronic painful knee osteoarthritis (OA). Methods. A parallel, two-arm, double-blind, randomized, sham-controlled clinical trial was conducted. People with chronic knee pain associated with OA were randomized to receive sham NF training or source-localized ratio ISF-NF training protocol to down-train ISF bands at the somatosensory (SSC), dorsal anterior cingulate (dACC), and uptrain pregenual anterior cingulate cortices (pgACC). Resting state EEG was recorded at baseline and immediate post-training. Results. The source localization mapping demonstrated a reduction (P = .04) in the ISF band activity at the left dorsolateral prefrontal cortex (LdlPFC) in the active NF group. Region of interest analysis yielded significant differences for ISF (P = .008), slow (P = .007), beta (P = .043), and gamma (P = .012) band activities at LdlPFC, dACC, and bilateral SSC. The FC between pgACC and left SSC in the delta band was negatively correlated with pain bothersomeness in the ISF-NF group. Conclusion. The EEG ISF-NF training can modulate EEG activity and connectivity in individuals with chronic painful knee osteoarthritis, and the observed EEG changes correlate with clinical pain measures.

Keywords

brain-computer interface neurofeedback infraslow slow cortical potentials knee osteoarthritis chronic pain neuromodulation

Introduction

Pain as a unified percept is proposed to be related to an imbalance in pathways in the central nervous system: the lateral “painfulness” pathway, the medial “suffering” pathway, and the descending pain “inhibitory” pathway.^1–5 The lateral pathway has the somatosensory cortex (SSC), the medial pathway has the dorsal anterior cingulate cortex (dACC), and the descending pain inhibitory pathway has the pregenual anterior cingulate cortex (pgACC) as their main cortical hubs.^3,4,6 Whereas pain was initially interpreted as activity changes in specific brain areas,⁷ it has become clear that pain may be an emergent property of pain networks in the brain, characterized by altered functional connectivity (FC)^3,8,9 or a combination of activity and FC changes. The FC permits communication between different brain areas involved in pain-evoking (pain severity and its associated suffering) and pain-suppressing (descending inhibition of nociceptive inputs) areas. Thus, a balance of pain evoking and suppression may determine the pain perceived.^3,4,6 Indeed, the pain has been shown to result from an imbalance between the activity in the dACC and SSC (pain evoking) and the pgACC (pain suppression).^3,4,6,10,11 Furthermore, the bigger the imbalance between pain input and pain suppression, the more intense the pain is perceived.⁶ Thus, modulating this imbalance should result in pain suppression and relief.

Non-invasive neuromodulatory approaches have been proposed as a potential intervention to modulate activity and FC between pain-processing brain regions with the potential to improve clinical pain outcomes.^12,13 Electroencephalography (EEG) based brain–computer interface (BCI) training, also known as EEG neurofeedback (EEG-NF), is a form of intervention developed for augmenting or reducing brain activity patterns that are linked to disease states.^13–15 Source-localized EEG-NF protocols can be designed to increase or decrease the intracortical neuronal activity at the targeted cortical networks or predefined regions of interest (ROIs) at a source level within the brain rather than only at a sensor level.^16,17 A large body of literature is available from various EEG-NF protocols tested for different pain conditions, including both RCTs^18–26 and non-RCTs.^27–37 Those EEG-NF protocols upregulate or downregulate higher frequency oscillations [alpha (8-12 Hz), beta (13-30 Hz), and gamma (30.5-44 Hz)] at a single or two ROIs. While these studies were able to improve clinical pain outcomes, a recent systematic review and meta-analysis demonstrated inconsistencies across these studies in reporting measures of brain changes. Most studies failed to show frequency-specific EEG changes in the trained or other frequencies within the spectrum.³⁸

Alterations in the FC between pain-evoking and pain-inhibiting regions have been demonstrated in various chronic pain populations.^39–44 It has been proposed that changes in the FC strength could be an objective marker of intervention efficacy alongside the changes reported in the subjective clinical measures.⁴¹ Thus far, only a few studies have investigated FC changes in pain following EEG-NF training and demonstrated FC changes at the sensor level⁴⁵ and specific ROIs.⁴⁶ However, none of the studies thus far have investigated source-localized region-specific FC changes and their association with pain and physical function measures.

Although the findings from existing EEG-NF indicate positive results with pain modulation, increasing evidence implicates the role of infraslow fluctuations (ISF–frequency < 0.1 Hz) in modulating dynamic brain connections.⁴⁷ ISF underlies the excitability of cortical networks and is robustly correlated with the phase of the faster frequency oscillations.⁴⁸ Moreover, higher frequency bands are nested (phase-locked) on the ISF bands (0.0-0.1 Hz), synchronizing the higher frequency activity of the brain.^49–53 Preclinical and empirical research shows neuropathic pain is associated with altered ISF within the dorsal horn, extending to the SSC.^54,55 Additionally, evidence from neuroimaging studies demonstrates increased ISF activity within the pain-evoking brain regions, including dACC, and SSC, and decreased ISF activity within the antinociceptive network, including pgACC.^56–66

Preliminary evidence suggests that training the ISF activity in pain processing regions can improve clinical outcomes such as pain and function.^65–70 Yet, it remains unclear whether ISF training can alter activity within the ISF and higher EEG bands. Furthermore, it is unknown whether such shifts in EEG activity correlate with changes in clinical outcomes. An exploratory, secondary analysis was undertaken to examine the effects of ISF-NF on EEG measures in people with chronic painful knee osteoarthritis (OA). Therefore, the primary objective was to determine whether ISF-NF training can result in measurable changes in brain activity in the full-band EEG spectrum [ISF (0.01-0.1 Hz), slow (0.2-1.5 Hz), delta (2-3.5 Hz), theta (4-7.5 Hz), alpha (8-12 Hz), beta (13-30 Hz), gamma (30.5-44 Hz)]. The secondary objectives include whether the ratio ISF-NF training protocol can alter the strength of functional connectivities between the targeted ROIs and to test whether the changes in EEG measures were correlated with changes in pain and physical function outcomes.

Methods

Design

This study was designed as a two-arm, parallel-group, double-blinded (participants and outcome assessor) randomized sham-controlled feasibility clinical trial.⁷⁰ Methodological descriptions of this study followed the Consolidated Standards of Reporting Trials (CONSORT) extension for randomized pilot and feasibility trials.⁷¹ The description of the study intervention was structured following the Template for Intervention Description and Replication (TIDieR) guide.⁷² This study was approved by the Health & Disability Ethics Committee (HDEC), New Zealand (19CEN182), and the trial was registered in the Australian New Zealand Clinical Trials Registry (ACTRN12620000273987). The research was conducted in accordance with the Declaration of Helsinki. A detailed version of the study protocol and the feasibility results were published elsewhere.^71,73

Participant Characteristics

Adults aged 44 to 75 years with a clinical diagnosis of knee OA; with pain (at least ≥ 4 on an 11-point numerical rating scale) for a minimum duration of 3 months were eligible to participate in the study. Participants were excluded if they had one of the following situations or conditions: (1) underwent surgery or other invasive procedures in the last six months and any surgical procedures scheduled within eight weeks after screening; (2) undertaken any steroid injections to the knee joint in the past three months or on oral steroids in the previous month; (3) current intake of centrally acting medications (eg, antidepressants, anticonvulsants, and neuropathic pain drugs) or intention of taking new medications in the next eight weeks; (4) neurological conditions or diseases (brain, spinal cord, or peripheral nerve injuries, radiculopathy, and neuropathies); (5) soft tissue injuries of the knee (eg, meniscus, muscle, tendon, or ligament injury) in the last three months; (6) cognitive impairments (dementia, posttraumatic stress disorders, and Alzheimer's disease); (7) difficulty or inability to read or understand English, or provide informed consent; and (8) pregnancy or six months postpartum.

Randomization and Allocation Concealment

Participants were randomized into either ISF-NF (active group) or sham NF (sham group) group using an open-access block randomization program by the department research administrator not involved in the assessments, allocation, or interventions. The allocation was concealed until after the initial assessment was completed. Participants and the outcome assessor were blinded to the group allocation. The flow of participants through the study phases is illustrated in Figure 1.^70,73

Figure 1.

The flow of participants through the study phases.

Assessments and Outcome Measures

EEG acquisition: Resting-state EEG (raw data with a sampling rate of 500 Hz) was obtained in a quiet room by an independent researcher. EEG was collected using the Mitsar EEG system with WinEEG software at T0 and T1.⁷⁴ The EEG was sampled with 19 Ag/AgCl electrodes (Fp1, Fp2, F7, F3, Fz, F4, F8, T7, C3, Cz, C4, T8, P7, P3, Pz, P4, P8, O1 O2) in the standard International 10-20 system placement referenced to linked ears and impedances was checked to remain below 5 kΩ.^75,76 Continuous EEG data was collected for 10 min with the participant's eyes closed. The participants’ alertness was assessed by monitoring both the deceleration of the alpha rhythm and the emergence of spindles in the EEG stream, aiming to avoid potential amplification of theta power caused by drowsiness during recording.^66,70,77

Clinical outcomes: The pain and function constructs were collected using validated questionnaires at T0 and T1. The multidimensional constructs were chosen based on the biopsychosocial model of pain literature.⁷⁸ Details, including psychometric properties and implementation method of each measure, are detailed in the published protocol.^71,73 A summary of the outcome measures is given in Table 1.

Table 1.

Summary of the Clinical Outcome Measures.

Questionnaires/measures	Components/dimensions/items	Scoring criteria/interpretation
Brief pain inventory (BPI)^79,80	24 h and 4 weeks pain severity items (worst pain, average pain, least pain, and current pain)	0—“No pain” to 10—“pain as bad as you can imagine”
	Mean pain severity subscore	Composite item taking the mean of the four 24-hour severity items
	Seven interference items (how pain interferes with activity, mood, relations with others, walking ability, work, enjoyment of life, and sleep)	0—“does not interfere” to 10—“completely interferes”
VAS-unpleasantness scale^81–83	Pain unpleasantness in the last 24 hours	0—being “not at all pleasant” and 15—being “most unpleasant imaginable”
Pain bothersomeness^84,85	Pain bothersomeness in the last week	Five possible responses: “not at all,” “slightly,” “moderately,” “very much,” and “extremely.” Scored 0 (not at all) to 4 (extremely)
Knee injury and osteoarthritis outcome score (KOOS)^86,87	42-item self-reported questionnaire.Five reported dimensions: pain (9 items), other symptoms (7 items), function in daily living (17 items), function in sport and recreation (5 items), and knee-related quality of life (4 items)	5-point Likert scale, with anchors of zero (no problems) to 4 (extreme problems). Scores are transformed to a 0- to 100 scale, with zero representing extreme knee problems and 100 representing no knee problems

Note: T0 = baseline; T1 = 2 days post-intervention.

EEG-Neurofeedback Protocol

Eligible participants were required to attend nine NF training sessions (30 min; 3 sessions/week) and two 90-min assessment sessions at T0 and T1. The experimental methodology and reporting of NF followed the consensus on the reporting and experimental design of clinical and cognitive-behavioral neurofeedback studies checklist (CRED-nf checklist-mandatory items).⁸⁸

During each session, participants were asked to sit on a chair with their backs supported and stay relaxed for 10 min, allowing the trainer to prepare them for the NF training. Both ISF-NF and sham-NF were administered using a 19-channel (ie, Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, and Pz) DC-coupled amplifier produced by BrainMaster Technologies, Inc.^{69,70,73,89–91} The impedance of the active electrodes was monitored and kept below 5 kΩ.⁶⁷ Participants were instructed to minimize eye movements, movements of the head and neck, swallowing, and teeth clenching to reduce motion artifacts in the EEG recordings. Additionally, they were informed that the sound they heard indicated successful performance.^92–97 Details of the intervention methodology and procedures can be found in previous publications.^70,73

Active ISF-NF ratio training protocol: A novel ISF-NF ratio training protocol was developed to enhance a balance between the activity of the three cortical areas [SSC, dACC, and pgACC].^3,4,6 Exact low-resolution electromagnetic tomography (eLORETA) ISF-NF was delivered using a DC-coupled amplifier produced by Brainmaster Inc. and the BrainAvatar software (Version 4.7.5.844). The ISF-NF training protocol (Figure 2) involved simultaneously downregulating the electrical activities of SSC (sensory-discriminative function) and dACC (motivational/affective function) and upregulating the pgACC (descending nociceptive inhibitory function).^{4,5,43,61,98–113}

Figure 2.

Regions of interest and ISF-NF training directions.

eLORETA source localization permits the selection of any region of the brain for feedback of the current density using voxels as ROI, which were selected based on the Montreal Neurological Institute (MNI) coordinate database,^114,115 previous literature^{3,4,6,69,70,116,117} and based on a Neurosynth meta-analysis (https://neurosynth.org/) of pain. A complete list of targeted voxels used for defining each ROI is available from Supplemental Materials 1 to 3. Figure 2 provides the center voxel coordinate for each ROI and the details of the ISF-NF training program.

The BrainMaster Technologies Software¹¹⁸ delivered sound feedback (reward) when the participant's brain activity met the desired ISF (0.0-0.1 Hz) threshold at the targeted ROIs via the laptop speakers. We utilized manual thresholding to adjust individual participant brain activity to meet the predetermined threshold if needed. This approach was selected based on insights from our prior study and the clinical experience of the authors.⁶⁷ Moreover, previous studies have reported better EEG learning with manual thresholding than with automated thresholding.^{93,95,119,120} The software calculated the ratio in real-time in the ISF band, and the feedback was given when the ratio was ≥ 1 based on satisfying the following equation:

\frac{2 \times pgACC}{SSC + dACC} \geq 1

Sham ISF-NF protocol: Conditions for the sham-NF group were the same as the ISF-NF group, except the participants received sound feedback recorded from someone else's prerecorded session. To ensure this, we have trained one healthy participant with the ISF-NF program for nine sessions. We captured the feedback sound using Audacity software,^121–123 a free and open-source digital audio editor and recording application. Participants in the sham-NF group were prepared in the same manner as those in the ISF-NF group, and they were provided with identical prerecorded feedback sounds. These prerecorded signals were randomly selected using the chit method from a pool of nine files.

Data Analysis

The EEG data were preprocessed using MATLAB (R2020a) with the EEGLAB toolbox^124,125 before the data were analyzed by eLORETA software. Each file was resampled to 128 Hz, bandpass filtered from 0.01 to 44 Hz, and EEG artifacts (eye blinks, muscle artifacts, perspiration, and body movements) were removed using IcoN Software Version 3.^6,67,69,126

The EEG data were analyzed using the open eLORETA method using LORETA-Key software to estimate the current density of intracerebral electrical sources that generate scalp-recorded electrical activity in each of the following seven frequency bands: ISF (0.01-0.1 Hz), slow (0.2-1.5 Hz), delta (2-3.5 Hz), theta (4-7.5 Hz), alpha (8-12 Hz), beta (13-30 Hz), and gamma (30.5-44 Hz).^127–129 We employed an average reference montage for localization.¹³⁰ eLORETA is one of the most robust methods for EEG source localization and can estimate current density over a grid of plausible cortical locations based on an inverse problem calculation.^{4,127,131–133}

eLORETA statistical contrast maps^67,134: eLORETA was used to generate voxel-based brain contrast maps of the cortical brain sources and frequency bands that are significantly different between the two groups using the highest threshold (Thresh) with P < .05. It is based on estimating, via randomization, the empirical probability distribution for the max statistic under the null hypothesis comparisons. The regions that demonstrated significant changes in the eLORETA mapping were also included for the ROI and the FC analysis in addition to the “a priori” ROIs (pgACC, dACC, S1Lt, and S1Rt).

Region of interest analysis: The log-transformed current density was extracted for all the ROIs (including the ones identified from the eLORETA statistical contrast maps) for the seven frequency bands (infraslow 0.01-0.1 Hz; slow 0.2-1.5 Hz; delta 2-3.5 Hz; theta 4-7.5 Hz; alpha 8-12 Hz; beta 12.5-30 Hz; and gamma 30.5-44 Hz) bands at T0 and T1 using the “LORETA to ROI” function in the eLORETA. To be consistent with the coordinates selected for the ISF-NF training and to minimize the source localization errors, all voxels within a radius of 5 mm were selected (a function in LORETA for defining the ROI) while extracting the current density for each ROI. The current density values were then imported to a customized Microsoft Excel sheet, and percentage change compared to baseline (relative current density) was calculated for both groups as follows:

Percent change to baseline = \frac{T 1 - T 0}{T 0} \times 100

ISF ratio computation: This study also computed the ISF band ratios between the NF-targeted ROIs using the following equation:

Left cortex : \frac{2 \times pgACC}{dACC + S 1 Lt}

Right cortex : \frac{2 \times pgACC}{dACC + S 1 Rt}

The final derived change scores were then exported to GraphPad Prism for statistical analysis. After conducting normality testing with the Shapiro–Wilk test and Q–Q plots, variances in relative current density among the groups across all ROIs and frequency bands were analyzed using the Mann-Whitney U test. We have included the regions that demonstrated significant changes in the source localized mapping for ROI and FC analysis, along with the NF-targeted ROIs.

Functional connectivity analysis: FC is a statistical measure of coherence, that is, co-varying activity or phase synchronization between two areas in the brain.^{4,6,116,135–138} FC consists of a combination of instantaneous and lagged phase synchronization. However, instantaneous phase synchronization is highly contaminated by the volume conduction of the electrical activity in the scalp.^4,106 The log-transformed lagged linear connectivity was derived for all the possible connections and frequency bands between the ROIs using the “connectivity” function in the eLORETA,¹³⁴ exported to a customized Microsoft Excel, and relative percentage change was computed following similar steps used in ROI analysis. A between-group analysis was performed using Mann-Whitney U test. The lagged phase synchronization between ROIs reflects the communication between those areas.¹¹⁵

Correlation between pain, physical function, and EEG measures: Based on the normality test (Shapiro–Wilk test and Q–Q plots), we performed a nonparametric Spearman correlation testing the statistical correlation between the change scores of pain (intensity, unpleasantness, bothersomeness, and interference), physical function, and EEG measures (ROI and FC).

Statistical Approach

The data processing and pooling for this study were done using either MATLAB 2020a^124,125 and/or Microsoft Excel (2020).¹³⁹ All the graphs and statistical analysis were derived using GraphPad Prism software version 9.1.0 for Windows (GraphPad Software, San Diego, CA, USA).¹⁴⁰ The normality of the data distribution for all the variables was confirmed using the Shapiro–Wilk test and Q–Q plots. Based on the outcome, nonparametric Mann-Whitney U test was used to compare the distribution of the two groups. The P-values were adjusted and stringently controlled for multiple comparisons using the Holm-Šídák test. The Holm-Šídák test is a powerful method for performing many comparisons using independent nonparametric statistics.^141,142 A P-value of 5% was used to accept statistically significant differences between the two groups. The GraphPad P-value reporting style was used, which reports four digits after the decimal point with a leading zero (0.1234). P values < .0001 are shown as “< .0001.” P values < .001 are summarized with three asterisks, and P values < .0001 are summarized with four asterisks.¹⁴³ Additionally, Cohen's d-effect size was calculated for significant between-group analyses.

Due to multiple ROIs and frequency bands in the analysis, a “Volcano plot” was created demonstrating the statistical summary for both ROI and FC analysis with Mann-Whitney mean rank difference on the X-axis and −log10(P-value) on the Y-axis. GraphPad Prism software automatically creates a volcano plot while performing multiple comparisons.¹⁴⁴ Volcano plot is a scatterplot used to demonstrate the statistical significance (P-value) versus the magnitude of change. The plot provides a quick visual understanding of the results based on the performed statistics.

Results

Demographics

Twenty-one participants with knee OA underwent baseline assessment and were randomly assigned to receive either ISF-NF (n = 11) or sham-NF (n = 10) group. Participants were middle-aged (61.7 ± 7.6 years), New Zealand European (90.5%), and females (62%) with an average knee pain duration of 4.0 ± 3.4 years. The socio-demographic and clinical characteristics of the participants in both groups and the results of tests comparing averages and proportions for the two groups at baseline are summarized in Table 2. Between-group comparisons revealed that both groups were comparable in age and sex, and no significant differences were found between groups.

Table 2.

Demographic and Clinical Characteristics of the Participants.

Characteristics	Active group (n = 11)	Sham group (n = 10)	P-value
Age, years, M ± SD	62.3 ± 8.5	61.0 ± 6.7	.67
Sex, n (%) female	7 (64%)	6 (60%)	—
Race, n (%)			—
New Zealand European	10 (91%)	9 (90%)	—
Australian	1 (9%)	0	—
Tongan	0	1 (10%)	—
Body Mass Index, kg/m², M ± SD	32.4 ± 9.5	30.4 ± 8.8	.63
Dominance, n (%) right	11 (100%)	10 (100%)	—
Average pain in the last 3 months (on NPRS)	6.1 ± 1.5	5.9 ± 1.2	.75
Affected knee, n (%)			—
Right	8 (72.7%)	7 (70%)	—
Left	3 (27.3%)	3 (30%)	—
Knee pain duration—years (M, SD)	5.3 ± 4	2.6 ± 2.3	.075
Current medication/treatment status			—
Analgesics	4 (36.4%)	6 (60%)	—
Physiotherapy	1 (9.1%)	1 (10%)	—
Analgesics and physiotherapy	2 (18.2%)	0	—
No-treatment	4 (36.4%)	3 (30%)	—

Electroencephalography Measures

eLORETA statistical contrast mapping analysis: The source localized mapping comparing both groups (independent group test) demonstrated a significant (P = .04, Thresh = −4.65, Cohen's d = 0.5) reduction in ISF band activity over the left dorsolateral prefrontal cortex (LdlPFC) in the ISF-NF group when compared to Sham-NF group (Figure 3).

Figure 3.

eLORETA statistical contrast maps: this figure is generated by the eLORETA after performing statistical non-parametric mapping on brain sources. The analysis demonstrates decreased activity in the ISF band in the LdlPFC (MNIxyz = −35, 40, 40).

We have included LdlPC as an ROI along with the “a priori” regions for further analyses (ROI and FC). The “a priori” ROIs in the present study are SSC left (S1Lt) (MNI coordinates [MNIxyz] = 18, −40, 68), SSC right (S1Rt) (MNIxyz = −18, −40, 68), pgACC (MNIxyz = 0, 36, 2), and dACC (MNIxyz = 0, 16, 32). Due to the midline proximity of dACC and pgACC, and considering the influence of volume conduction, a laterality differentiation for these areas was not done.⁶

ROI analysis: The Mann-Whitney U test yielded a significant difference for the ISF band activity in the LdlPFC (P = .008, Cohen's d = 1.2), slow band in the dACC (P = .0079, Cohen's d = 0.83), gamma band in S1Lt (P = .0127, Cohen's d = 0.77) and S1Rt (P = .0242, Cohen's d = 0.82), and beta in S1Lt (P = .0430, Cohen's d = 0.57). Figure 4A to H demonstrates the graphical summary of the analysis. No other significant differences were found for ROIs and frequency bands.

Figure 4.

Displays the outcomes of the region of interest (ROI) analysis conducted with the Mann-Whitney U test: (A-G) Demonstrates the average current density of the percentage change to baseline (means with 95% CI) for each ROI and frequency band for both groups. (H) Provides a statistical summary of the ROI analysis based on the mean rank difference and P values. Variables with lesser P-values (higher significance) appear on the top upper panel (blue color). Data points below the center X-bar (at P = .05) are nonsignificant changes.

The study found no significant change in the ISF ratio for both the right and left cortical regions over the NF-targeted regions (Figure 5).

Figure 5.

ISF ratio analysis: the box and whiskers plot demonstrates (means with 95% CI) that the ISF ratio changes for both groups.

Functional Connectivity (Lagged Phase Synchronization)

The lagged linear connectivity analysis between groups revealed significant functional connectivity changes (Figure 6) in the slow band (S1Lt-LdlPFC, P = .042, Cohen's d = 0.93), delta (pgACC-S1Lt, P = .029, Cohen's d = 0.65; dACC-S1Lt, P = .042, Cohen's d = 0.57), theta (S1Rt-LdlPFC, P = .015, Cohen's d = 1.03; dACC-LdlPFC, P = .029, Cohen's d = 0.70), and beta (pgACC-LdlPFC, P = .0036, Cohen's d = 1.4; pgACC-dACC, P = .0196, Cohen's d = 1.2) more pronounced in the sham-NF group. No significant changes were found for other frequencies and connections.

Figure 6.

The results of functional connectivity analysis using the Mann-Whitney U test: (A-G) demonstrate functional connectivity strength between regions of interest (ROIs) for the seven frequency bands of interest. (H) Provides a statistical summary of the FC analysis based on the mean rank difference and P values. Variables with lesser P-values (higher significance) appear on the top upper panel (blue). Data points below the center X-bar (at P = .05) are nonsignificant changes.

Correlation Between Pain, Physical Function, and EEG Measures

The Spearman correlation yielded a moderate positive correlation between ISF band activity in the LdlPFC and pain unpleasantness (ρ = 0.6698, P = .028) in the active group. Also, the FC between pgACC and S1Lt in the delta band demonstrated a moderate negative correlation (ρ = −0.626, P = .04) with pain bothersomeness in the active group.

No other significant correlations were found for other measures in both groups.

Discussion

Principal Findings

This study sought to determine whether the ISF-NF ratio training can result in measurable brain activity and FC changes in individuals suffering from chronic knee OA pain. The between-group source localization mapping demonstrated that the ISF-NF training significantly decreased activity at the left dlPFC compared to the sham group. ROI analysis also revealed a significant reduction of the ISF current density in the LdlPFC in the active group compared to the control group. Moreover, the decreased ISF power in the LdlPFC in the active group was positively correlated with the pain unpleasantness measure. FC changes were observed in the slow, delta, theta, and beta bands between various ROIs in both groups. Notably, the FC between pgACC and S1Lt in the delta band was negatively correlated with the pain bothersomeness change score in the ISF-NF group.

eLORETA Statistical Contrast Maps and Region of Interest Analysis

The eLORETA statistical contrast mapping method makes no “a priori” assumption about the brain regions. Instead, it performs a source-localized statistical parametric mapping to identify brain regions with significant changes compared to a control group.^145,146 In our study, LdlPFC is the only region that demonstrated a significant change in the mapping with decreased ISF activity of the region in the active group. Additionally, the ROI analysis also showed a significant reduction of the ISF activity in the LdlPFC in the active group. The dlPFC is an important heterogeneous cortical region that mediates the antinociceptive, cognitive, and affective processing of pain.^147–149 Importantly, dlPFC is a key node of several brain networks primarily implicated in the cognitive-affective and sensory processing of pain experience.^{147,148,150,151} Moreover, the dlPFC is functionally connected and positively correlated with the activity of the dACC, another important cortical hub modulating the affective dimensions of the pain experience.^{3,116,152,153} Previous studies have reported reduced functional and structural activity of the PFC in various chronic pain populations^154–158 and have reported successful noninvasive stimulation of the PFC for treating chronic pain.^45,159,160 In this study, we found no significant change in ISF neural activity over the targeted regions in both eLORETA brain mapping and ROI analysis. However, the training impacted the regions that are functionally connected with the targeted regions. The ratio ISF training down-regulated dACC may have resulted in the reduced ISF activity of the dlPFC in the active group. This evidence suggests an indirect effect of ISF-NF on other regions involved in the same functional network. This is of no surprise considering the profound role of ISF in modulating and synchronizing various cortical networks of the brain.^50–53 In addition, the ISF activity of the LdlPFC was positively correlated with the unpleasantness of pain in the active group. Previous studies have described dACC and dlPFC as a region mediating unpleasant and affective aspects of pain.^{1,5,103,153,161,162} This is an important finding and supports the claim that the displayed ISF effects in the dlPFC are likely to be modulated by the downregulation of the dACC, thus improving clinical symptomatology linked to the regions.

However, it is worth mentioning that the training protocol was designed to increase the ISF activity of pgACC and simultaneously downregulate dACC and SSC (S1Lt and S1Rt). An unexpected but important finding was that we did not observe any activity changes in the targeted regions in the direction of training for the ISF bands. Previous studies in EEG-NF have also reported similar observations in tinnitus-related distress and cognitive NF training.^76,163 The authors speculate that this could be due to the training of a relatively smaller cortical region to influence pain-related networks, where several cortical and subcortical circuits operate to modulate pain experience.⁶² This could also explain why we could not find any significant changes in the ISF ratio. Additionally, the spatial specificity in localizing the targeted region using a 19-electrode head cap can be influenced by the activity of the nearby regions close to the targeted region, which could also have influenced the results.^163,164 However, this needs to be investigated with an extended protocol influencing multiple brain regions using high-resolution EEG with more number of electrodes.¹⁶⁵ In summary, even though a ratio of activity (current density) was trained, only changes in FC are found and not limited to the trained frequency band.

Functional Connectivity Changes and Potential Clinical Benefits of Ratio ISF-NF Training

In chronic pain, decreased communication (ie, FC) has been demonstrated between pgACC and SSC.⁶ Moreover, FC alterations are observed in chronic pain in the regions or associated networks tested in this investigation.^166–170 Supporting the literature, the study demonstrated FC changes between ROIs in the slow band (S1Lt-LdlPFC), delta (pgACC-S1Lt; dACC-S1Lt), theta (S1Rt-LdlPFC; dACC-LdlPFC), and beta (pgACC-LdlPFC; pgACC-dACC). Most of these changes were also observed in the sham group. However, the FC changes exhibited a negative correlation between the FC of pgACC and S1Lt in the delta band, with pain bothersomeness only in the active group. One of the ISF-NF ratio components was to uptrain the ISF activity of the pgACC. We believe the training must have influenced the functional communication from pgACC to SSC, thus influencing the clinical pain outcomes.

Limitations and Future Research Recommendations

Given that this is the first ISF-NF ratio study, all the analyses performed were explorative, informing the importance of a full clinical trial. We recruited a small sample of individuals with knee OA with a homogeneous sex and race distribution, which restricts the generalizability of the study findings. Furthermore, all correlation analyses conducted in the study were exploratory. Consequently, the study refrained from applying corrections to the p values for multiple bivariate correlations. However, it is important to acknowledge that employing the Bonferroni correction method might enhance the risk of type II errors, potentially overlooking real relationships.¹⁷¹ Despite the limitations, this study provides an essential foundation for designing and implementing future clinical trials investigating ISF-NF protocols.

Ratio training protocols are more functional and reduce the hassle of separately training individual ROIs. However, further investigations are required to determine the efficiency of the ratio training protocol against the training of cortical areas separately on EEG measures at prespecified ROI and network-level changes. Similarly, it is also worth investigating whether NF training of higher EEG bands can influence the ISF band and comparing the effectiveness of ISF-NF against higher band NF training protocols on clinical outcomes in people with chronic pain.

Conclusions

The ratio ISF-NF training can result in measurable changes in brain activity and functional connectivity in individuals suffering from chronic knee OA pain. However, ISF-NF training did not change the EEG activity of the regions exposed to training; instead, it exerted its effect on a functional network level. Caution is warranted in interpreting these findings since it is explorative analyses on an unpowered sample, and the observations could be due to chance.

Supplemental Material

sj-csv-1-eeg-10.1177_15500594241264892 - Supplemental material for Closed-Loop Infraslow Brain–Computer Interface can Modulate Cortical Activity and Connectivity in Individuals With Chronic Painful Knee Osteoarthritis: A Secondary Analysis of a Randomized Placebo-Controlled Clinical Trial

Supplemental material, sj-csv-1-eeg-10.1177_15500594241264892 for Closed-Loop Infraslow Brain–Computer Interface can Modulate Cortical Activity and Connectivity in Individuals With Chronic Painful Knee Osteoarthritis: A Secondary Analysis of a Randomized Placebo-Controlled Clinical Trial by Jerin Mathew, Divya Bharatkumar Adhia, Mark Llewellyn Smith, Dirk De Ridder and Ramakrishnan Mani in Clinical EEG and Neuroscience

Supplemental Material

sj-csv-2-eeg-10.1177_15500594241264892 - Supplemental material for Closed-Loop Infraslow Brain–Computer Interface can Modulate Cortical Activity and Connectivity in Individuals With Chronic Painful Knee Osteoarthritis: A Secondary Analysis of a Randomized Placebo-Controlled Clinical Trial

Supplemental material, sj-csv-2-eeg-10.1177_15500594241264892 for Closed-Loop Infraslow Brain–Computer Interface can Modulate Cortical Activity and Connectivity in Individuals With Chronic Painful Knee Osteoarthritis: A Secondary Analysis of a Randomized Placebo-Controlled Clinical Trial by Jerin Mathew, Divya Bharatkumar Adhia, Mark Llewellyn Smith, Dirk De Ridder and Ramakrishnan Mani in Clinical EEG and Neuroscience

Supplemental Material

sj-csv-3-eeg-10.1177_15500594241264892 - Supplemental material for Closed-Loop Infraslow Brain–Computer Interface can Modulate Cortical Activity and Connectivity in Individuals With Chronic Painful Knee Osteoarthritis: A Secondary Analysis of a Randomized Placebo-Controlled Clinical Trial

Supplemental material, sj-csv-3-eeg-10.1177_15500594241264892 for Closed-Loop Infraslow Brain–Computer Interface can Modulate Cortical Activity and Connectivity in Individuals With Chronic Painful Knee Osteoarthritis: A Secondary Analysis of a Randomized Placebo-Controlled Clinical Trial by Jerin Mathew, Divya Bharatkumar Adhia, Mark Llewellyn Smith, Dirk De Ridder and Ramakrishnan Mani in Clinical EEG and Neuroscience

Footnotes

Acknowledgments

We acknowledge the International Society for Neuroregulation and Research (ISNR) and the International QEEG Certification Board for supporting the first author through monetary awards and scholarships in support of this study. The authors would like to thank Hazel Potterton, Tyson Michael Perez, and Matt Hall for their active support and help throughout this study. Also, the authors thank all the study participants for their time and support in completing this study during the COVID-19 pandemic in 2020.

Data Availability Statement

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We acknowledge the International Society for Neuroregulation and Research (ISNR) and the International QEEG Certification Board for supporting the first author through monetary awards and scholarships in support of this study. This study was also supported by the Centre for Health, Activity, and Rehabilitation Research (CHARR) yearly graduate research fund.

Ethical Statement

This study was approved by the Health & Disability Ethics Committee (HDEC), New Zealand (19CEN182), and the trial was registered in the Australian New Zealand Clinical Trials Registry (ACTRN12620000273987).

ORCID iD

Jerin Mathew

Supplemental material

Supplemental material for this article is available online.

References

Price

. Psychological and neural mechanisms of the affective dimension of pain. Science. 2000;288(5472):1769-1772.

Fields

. State-dependent opioid control of pain. Nat Rev Neurosci. 2004;5(7):565-575.

De Ridder

Adhia

Vanneste

. The anatomy of pain and suffering in the brain and its clinical implications. Neurosci Biobehav Rev. 2021;130:125-146.

De Ridder

Vanneste

. The Bayesian brain in imbalance: medial, lateral and descending pathways in tinnitus and pain: a perspective. Prog Brain Res. 2021;262:309-334.

Bushnell

Ceko

Low

. Cognitive and emotional control of pain and its disruption in chronic pain. Nat Rev Neurosci. 2013;14(7):502-511.

Vanneste

De Ridder

. Chronic pain as a brain imbalance between pain input and pain suppression. Brain Commun. 2021;3(1): fcab014.

Peyron

Laurent

Garcia-Larrea

. Functional imaging of brain responses to pain. A review and meta-analysis (2000). Neurophysiol Clin. 2000;30(5):263-288.

Brandl

Weise

Mulej Bratec

, et al. Common and specific large-scale brain changes in major depressive disorder, anxiety disorders, and chronic pain: a transdiagnostic multimodal meta-analysis of structural and functional MRI studies. Neuropsychopharmacology. 2022;47:1071-1080.

Kolesar

Bilevicius

Kornelsen

. Salience, central executive, and sensorimotor network functional connectivity alterations in failed back surgery syndrome. Scand J Pain. 2017;16:10-14.

10.

De Ridder

Vanneste

. Burst and tonic spinal cord stimulation: different and common brain mechanisms. Neuromodulation. 2016;19(1):47-59.

11.

Mathew

Perez

Adhia

De Ridder

Mani

. Is there a difference in EEG characteristics in acute, chronic, and experimentally induced musculoskeletal pain states? a systematic review. Clin EEG Neurosci. 2024;.55(1):101-120.

12.

Meeker

Jupudi

Lenz

Greenspan

. New developments in non-invasive brain stimulation in chronic pain. Curr Phys Med Rehabil Rep. 2020;8(3):280-292.

13.

Roy

de la Vega

Jensen

Miro

. Neurofeedback for pain management: a systematic review. Front Neurosci. 2020;14(671).

14.

Enriquez-Geppert

Huster

Herrmann

. EEG-neurofeedback as a tool to modulate cognition and behavior: a review tutorial. Front Hum Neurosci. 2017;11(51).

15.

Sitaram

Ros

Stoeckel

, et al. Closed-loop brain training: the science of neurofeedback. Nat Rev Neurosci. 2017;18(2):86-100.

16.

Marzbani

Marateb

Mansourian

. Neurofeedback: a comprehensive review on system design, methodology and clinical applications. Basic Clin Neurosci. 2016;7(2):143-158.

17.

Congedo

Lubar

Joffe

. Low-resolution electromagnetic tomography neurofeedback. IEEE Trans Neural Syst Rehabil Eng. 2004;12(4):387-397.

18.

Cohen

Mcarthur

Rickles

. Comparison of four biofeedback treatments for migraine headache: psychological and headache variables. Psychosom Med. 1980;42(5):463-480.

19.

Mathew

Mishra

Kumaraiah

. Alpha feedback in the treatment of tension headache. J of Personal Clin Stud. 1987.

20.

Kravitz

Fawcett

Esty

Katz

. Treatment of fibromyalgia syndrome using low-intensity neurofeedback with the flexyx neurotherapy system: a randomized controlled clinical trial. J Neurother. 2006;10(2-3):41-58. doi:https://doi.org/10.1300/J184v10n02_03

21.

Kayiran

. Neurofeedback intervention in fibromyalgia syndrome; a randomized, controlled, rater blind clinical trial. Appl Psychophysiol Biofeedback. 2010;35(4):293.

22.

Nelson

Bennett

Barkhuizen

, et al.

Neurotherapy of fibromyalgia?

Pain Medicine (Malden, Mass). 2010;11(6):912.

23.

Farahani

Tavallaie

Ashtiani

. Comparison of neurofeedback and transcutaneous electrical nerve stimulation efficacy on treatment of primary headaches: a randomized controlled clinical trial. Iran Red Crescent Med J. 2014;16(7):e17799.

24.

Prinsloo

Novy

Driver

, et al. The long-term impact of neurofeedback on symptom burden and interference in patients with chronic chemotherapy-induced neuropathy: analysis of a randomized controlled trial. J Pain Symptom Manage. 2018; 55(5):1276-1285.

25.

Y-L

Fang

S-C

Chen

S-C

Tai

C-J

Tsai

P-S

. Effects of neurofeedback on fibromyalgia: a randomized controlled trial. Pain Manage Nurs. 2021.

26.

Goldway

Ablin

Lubin

, et al. Volitional limbic neuromodulation exerts a beneficial clinical effect on fibromyalgia. Neuroimage. 2019;186(758).

27.

Kayiran

Dursun

Ermutlu

Dursun

Karamürsel

. Neurofeedback in fibromyalgia syndrome. Agri: Agri (Algoloji) Dernegi'nin Yayin Organidir = The Journal of the Turkish Society of Algology. 2007;19(3):47-53.

28.

Jensen

Grierson

Tracy-Smith

Bacigalupi

Othmer

. Neurofeedback treatment for pain associated with complex regional pain syndrome type I. J Neurother. 2007;11(1):45-53.

29.

Ibric

Dragomirescu

. Neurofeedback in pain management. In Introduction to quantitative EEG and neurofeedback: advanced theory and applications. 2nd ed. Elsevier; 2009:417-451.

30.

Winter

Caro

. EEG Biofeedback treatment improves certain attention and somatic symptoms in fibromyalgia: a pilot study. Appl Psychophysiol Biofeedback. 2011;36(3):193-200.

31.

Jensen

Gertz

Kupper

, et al. Steps toward developing an EEG biofeedback treatment for chronic pain. Appl Psychophysiol Biofeedback. 2013;38(2):101-108.

32.

Jensen

Askew

Gianas

, et al. Effects of non-pharmacological pain treatments on brain states. Clin Neurophysiol. 2013;124(10):2016-2024.

33.

Nelson

Esty

. Neurotherapy for chronic headache following traumatic brain injury. Mil Med Res. 2015;2:22. Headache. 2016;56(2):411.

34.

Jacobs

Jensen

. EEG neurofeedback in the treatment of chronic pain: a case series. NeuroRegulation. 2015;2(2):86-86.

35.

Mayaud

Barthélemy

, et al. Alpha-phase synchrony EEG training for multi-resistant chronic low back pain patients: an open-label pilot study. Eur Spine J. 2019;28(11):2487-2501.

36.

Petric-Gray

Vuckovic

Al-Taleb

MKH

Purcell

Fraser

. Home used, patient self-managed, brain-computer interface for the management of central neuropathic pain post spinal cord injury: usability study. J Neuroeng Rehabil. 2019;16(1):128.

37.

Hershaw

Hill-Pearson

Arango

Souvignier

Pazdan

. Semi-automated neurofeedback therapy for persistent postconcussive symptoms in a military clinical setting: a feasibility study. Mil Med. 2020;185(3-4):e457-e465.

38.

Hesam-Shariati

Chang

Wewege

, et al. The analgesic effect of electroencephalographic neurofeedback for people with chronic pain: a systematic review and meta-analysis. Eur J Neurol. 2022;29(3):921-936.

39.

Motoyama

Oshiro

Takao

, et al. Resting-state brain functional connectivity in patients with chronic pain who responded to subanesthetic-dose ketamine. Sci Rep. 2019;9(1):12912.

40.

Necka

Lee

I-S

Kucyi

Cheng

Atlas

. Applications of dynamic functional connectivity to pain and its modulation. Pain Reports. 2019;4(4):e752-e752.

41.

Thorp

Suchy

Vadivelu

Helander

Urman

Kaye

. Functional connectivity alterations: novel therapy and future implications in chronic pain management. Pain Physician. 2018;21(3):e207-e214.

42.

Kaplan

Schrepf

Vatansever

, et al. Functional and neurochemical disruptions of brain hub topology in chronic pain. PAIN. 2019;160(4):973-983.

43.

Spisak

Kincses

Schlitt

, et al. Pain-free resting-state functional brain connectivity predicts individual pain sensitivity. Nat Commun. 2020;11(1):187.

44.

Fallon

Chiu

Nurmikko

Stancak

. Functional connectivity with the default mode network is altered in fibromyalgia patients. PLoS One. 2016;11(7):e0159198.

45.

Hassan

Fraser

Conway

Allan

Vuckovic

. The mechanism of neurofeedback training for treatment of central neuropathic pain in paraplegia: a pilot study. BMC Neurol. 2015;15(1):200.

46.

Terrasa

Barros-Loscertales

Montoya

Muñoz

. Self-regulation of SMR power led to an enhancement of functional connectivity of somatomotor cortices in fibromyalgia patients. Original research. Front Neurosci. 2020.

47.

Ploner

Sorg

Gross

. Brain rhythms of pain. Trends Cogn Sci. 2017;21(2):100-110.

48.

Vanhatalo

Voipio

Kaila

. Full-band EEG (FbEEG): an emerging standard in electroencephalography. Clin Neurophysiol. 2005;116(1):1-8.

49.

Hyafil

Giraud

Fontolan

Gutkin

. Neural cross-frequency coupling: connecting architectures, mechanisms, and functions. Trends Neurosci. 2015;38(11):725-740.

50.

Dong

Liu

Wang

, et al. Low-frequency fluctuation in continuous real-time feedback of finger force: a new paradigm for sustained attention. Neurosci Bull. 2012;28(4):456-467.

51.

Mairena

Di Martino

Dominguez-Martin

, et al. Low frequency oscillations of response time explain parent ratings of inattention and hyperactivity/impulsivity. Eur Child Adolesc Psychiatry. 2012;21(2):101-109.

52.

Monto

Palva

Voipio

Palva

. Very slow EEG fluctuations predict the dynamics of stimulus detection and oscillation amplitudes in humans. J Neurosci. 2008;28(33):8268-8272.

53.

Smith

Collura

Ferrera

de Vries

. Infra-slow fluctuation training in clinical practice: a technical history. NeuroRegulation. 2014;1(2):187-187.

54.

Gerke

Duggan

Siddall

. Thalamic neuronal activity in rats with mechanical allodynia following contusive spinal cord injury. Neuroscience. 2003;117(3):715-722.

55.

Iwata

LeBlanc

Kadasi

Zerah

Cosgrove

Saab

. High-frequency stimulation in the ventral posterolateral thalamus reverses electrophysiologic changes and hyperalgesia in a rat model of peripheral neuropathic pain. Pain. 2011;152(11):2505-2513.

56.

Alshelh

Di Pietro

Youssef

, et al. Chronic neuropathic pain: it's about the rhythm. J Neurosci. 2016;36(3):1008-1018.

57.

Alshelh

. Infra-slow oscillations in chronic orofacial neuropathic pain and the effects of palmitoylethanolamide . Doctor of Philosophy PhD. University of Sydney Faculty of Medicine and Health Department of Anatomy and Histology Laboratory of Neural Imaging; 2018. http://hdl.handle.net/2123/18655

58.

Zhou

Hong

, et al. Altered low-frequency oscillation amplitude of resting state-fMRI in patients with discogenic low-back and leg pain. J Pain Res. 2018;11:165-176.

59.

Zhang

Jung

, et al. Identifying brain regions associated with the neuropathology of chronic low back pain: a resting-state amplitude of low-frequency fluctuation study. Br J Anaesth. 2019;123(2):e303-e311.

60.

Majeed

Magnuson

Hasenkamp

, et al. Spatiotemporal dynamics of low frequency BOLD fluctuations in rats and humans. Neuroimage. 2011;54(2):1140-1150.

61.

Kucyi

Salomons

Davis

. Mind wandering away from pain dynamically engages antinociceptive and default mode brain networks. Proc Natl Acad Sci U S A. 2013;110(46):18692-18697.

62.

Kucyi

Davis

. The dynamic pain connectome. Trends Neurosci. 2015;38(2):86-95.

63.

Di Pietro

Lee

Henderson

. Altered resting activity patterns and connectivity in individuals with complex regional pain syndrome. Hum Brain Mapp. 2020;41(13):3781-3793.

64.

Zhang

Mao

Pan

, et al. Frequency-specific alterations in cortical rhythms and functional connectivity in trigeminal neuralgia. Brain Imaging Behav. 2019;13(6):1497-1509.

65.

Adhia

Mani

Mathew

, et al. Exploring electroencephalographic infraslow neurofeedback treatment for chronic low back pain: a double-blinded safety and feasibility randomized placebo-controlled trial. Sci Rep. 2023;13(1):1177.

66.

Mathew

Adhia

Hall

De Ridder

Mani

. EEG-based cortical alterations in individuals with chronic knee pain secondary to osteoarthritis: a cross-sectional investigation. J Pain. 2024;25(5):104429.

67.

Leong

Vanneste

Lim

Smith

Manning

De Ridder

. A randomised, double-blind, placebo-controlled parallel trial of closed-loop infraslow brain training in food addiction. Sci Rep. 2018;8(1):11659.

68.

Balt

Du Toit

Smith

van Rensburg

. The effect of infraslow frequency neurofeedback on autonomic nervous system function in adults with anxiety and related diseases. NeuroRegulation. 2020;7(2):64-64.

69.

Adhia

Mani

Turner

Vanneste

De Ridder

. Infraslow neurofeedback training alters effective connectivity in individuals with chronic low back pain: a secondary analysis of a pilot randomized placebo-controlled study. Brain Sci. 2022;12(11):1514.

70.

Mathew

Adhia

Smith

De Ridder

Mani

. Source localized infraslow neurofeedback training in people with chronic painful knee osteoarthritis: a randomized, double-blind, sham-controlled feasibility clinical trial. Original research. Front Neurosci. 2022;16.

71.

Eldridge

Chan

Campbell

, et al. CONSORT 2010 statement: extension to randomised pilot and feasibility trials. BMJ. 2016;355:i5239.

72.

Hoffmann

Glasziou

Boutron

, et al. Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. BMJ. 2014;348(g1687).

73.

Mathew

Adhia

Smith

De Ridder

Mani

. Protocol for a pilot randomized sham-controlled clinical trial evaluating the feasibility. Safety, and acceptability of infraslow electroencephalography neurofeedback training on experimental and clinical pain outcomes in people with chronic painful knee. NeuroRegulation. 2020;7(1):30-44.

74.

Gorecka

Makiewicz

. The dependence of electrode impedance on the number of performed EEG examinations. Sensors (Basel). 2019;19(11):2608.

75.

Khazi

Kumar

Vidya

. Analysis of EEG using 10: 20 electrode system. Int J of Innovati Res in Sci, Eng Technol. 2012;1(2):185-191.

76.

Vanneste

Joos

Ost

De Ridder

. Influencing connectivity and cross-frequency coupling by real-time source localized neurofeedback of the posterior cingulate cortex reduces tinnitus related distress. Neurobiol Stress. 2018;8:211-224.

77.

Britton

Frey

Hopp

, et al. Electroencephalography (EEG): An Introductory Text and Atlas of Normal and Abnormal Findings in Adults, Children, and Infants. American Epilepsy Society; 2016.

78.

Dworkin

Turk

Farrar

, et al. Core outcome measures for chronic pain clinical trials: IMMPACT recommendations. Pain (Amsterdam). 2005;113(1):9-19.

79.

Mendoza

Mayne

Rublee

Cleeland

. Reliability and validity of a modified brief pain inventory short form in patients with osteoarthritis. Eur J Pain. 2006;10(4):353-361.

80.

Keller

Bann

Dodd

Schein

Mendoza

Cleeland

. Validity of the brief pain inventory for use in documenting the outcomes of patients with noncancer pain. Clin J of Pain. 2004;20(5):309-318.

81.

Starr

Sawaki

Wittenberg

, et al. The contribution of the putamen to sensory aspects of pain: insights from structural connectivity and brain lesions. Brain. 2011;134(Pt 7):1987-2004.

82.

Price

Mcgrath

Rafii

Buckingham

. The validation of visual analog scales as ratio scale measures for chronic and experimental pain. Pain. 1983;17(1):45-56. doi:10.1016/0304-3959(83)90126-4

83.

Harkins

Price

Braith

. Effects of extraversion and neuroticism on experimental pain, clinical pain, and illness behavior. Pain. 1989;36(2):209-218.

84.

Dunn

Croft

. Classification of low back pain in primary care: using “bothersomeness” to identify the most severe cases. Spine. 2005;30(16):1887-1892.

85.

Thomas

Rathod-Mistry

Harper

, et al. Acute flares of knee osteoarthritis (the ACT-FLARE study): protocol for a web-based case-crossover study in community-dwelling adults. JMIR Res Protoc. 2019;8(4):e13428.

86.

Hawker

Davis

French

, et al. Development and preliminary psychometric testing of a new OA pain measure—an OARSI/OMERACT initiative. Osteoarthritis Cartilage. 2008;16(4):409-414.

87.

Peer

Lane

. The knee injury and osteoarthritis outcome score (KOOS): a review of its psychometric properties in people undergoing total knee arthroplasty. J Orthop Sports Phys Ther. 2013;43(1):20-28.

88.

Ros

Enriquez-Geppert

Zotev

, et al. Consensus on the reporting and experimental design of clinical and cognitive-behavioural neurofeedback studies (CRED-nf checklist). Oxford University Press; 2020.

89.

Perez

Glue

Adhia

, et al.. 2021. The ISAD (infraslow neurofeedback for anxiety and depression) study: a protocol for a trans-diagnostic, randomized, double-blind, sham-controlled, dose-response. Parallel-Group Trial. PREPRINT (Version 1) available at Research Square [https://doi.org/10.21203/rs.3.rs-173667/v1]

90.

Burkhard

Elmer

Kara

Brauchli

Jäncke

. The effect of background music on inhibitory functions: an ERP study. Front Hum Neurosci. 2018;12(293).

91.

Greber

Jäncke

. Suppression of pitch labeling: no evidence for an impact of absolute pitch on behavioral and neurophysiological measures of cognitive inhibition in an auditory go/no go task. Front Hum Neurosci. 2020;14(585505).

92.

Sepulveda

Sitaram

Rana

Montalba

Tejos

Ruiz

. How feedback, motor imagery, and reward influence brain self-regulation using real-time fMRI. Hum Brain Mapp. 2016;37(9):3153-3171.

93.

Strehl

. What learning theories can teach us in designing neurofeedback treatments. Front Hum Neurosci. 2014;8(894).

94.

Kober

Witte

Ninaus

Neuper

Wood

. Learning to modulate one's own brain activity: the effect of spontaneous mental strategies. Front Hum Neurosci. 2013;7(695).

95.

Micoulaud-Franchi

J-A

Mcgonigal

Lopez

Daudet

Kotwas

Bartolomei

. Electroencephalographic neurofeedback: level of evidence in mental and brain disorders and suggestions for good clinical practice. Neurophysiologie Clinique/Clin Neurophysiol. 2015;45(6):423-433.

96.

Gruzelier

. EEG-neurofeedback for optimising performance. III: A review of methodological and theoretical considerations. Neurosci Biobehav Rev. 2014;44:159-182.

97.

Sulzer

Haller

Scharnowski

, et al. Real-time fMRI neurofeedback: progress and challenges. Neuroimage. 2013;76:386-399.

98.

Fallon

Stancak

. Pain catastrophising affects cortical responses to viewing pain in others. PLoS One. 2015;10(7):e0133504.

99.

Kong

Loggia

Zyloney

LaViolette

Gollub

. Exploring the brain in pain: activations, deactivations and their relation. PAIN. 2010;148(2):257-267.

100.

Kwon

Altin

Duenas

Alev

. The role of descending inhibitory pathways on chronic pain modulation and clinical implications. Pain Pract. 2014;14(7):656-667.

101.

Ossipov

Morimura

Porreca

. Descending pain modulation and chronification of pain. Curr Opin Support Palliat Care. 2014;8(2):143-151.

102.

Rainville

. Brain mechanisms of pain affect and pain modulation. Curr Opin Neurobiol. 2002;12(2):195-204.

103.

Rainville

Duncan

Price

Carrier

Bushnell

. Pain affect encoded in human anterior cingulate but not somatosensory cortex. Science. 1997;277(5328):968-971.

104.

Vogt

Sikes

. The medial pain system, cingulate cortex, and parallel processing of nociceptive information. Prog Brain Res. 2000;122:223-235.

105.

Yam

Loh

Tan

Khadijah Adam

Abdul Manan

Basir

. General pathways of pain sensation and the major neurotransmitters involved in pain regulation. Int J Mol Sci. 2018;19(8):2164.

106.

Vanneste

Ost

Van Havenbergh

De Ridder

. Resting state electrical brain activity and connectivity in fibromyalgia. PLoS One. 2017;12(6):e0178516.

107.

Baliki

Geha

Jabakhanji

Harden

Schnitzer

Apkarian

. A preliminary fMRI study of analgesic treatment in chronic back pain and knee osteoarthritis. Mol Pain. 2008;4(47).

108.

Baliki

Mansour

Baria

Apkarian

. Functional reorganization of the default mode network across chronic pain conditions. PLoS One. 2014;9(9):e106133.

109.

Cottam

Iwabuchi

Drabek

Reckziegel

Auer

. Altered connectivity of the right anterior insula drives the pain connectome changes in chronic knee osteoarthritis. Pain. 2018;159(5):929-938.

110.

Furman

Meeker

Rietschel

, et al. Cerebral peak alpha frequency predicts individual differences in pain sensitivity. Neuroimage. 2018;167:203-210.

111.

Howard

Sanders

Krause

, et al. Alterations in resting-state regional cerebral blood flow demonstrate ongoing pain in osteoarthritis: an arterial spin-labeled magnetic resonance imaging study. Arthritis Rheum. 2012;64(12):3936-3946.

112.

Kulkarni

Bentley

Elliott

, et al. Arthritic pain is processed in brain areas concerned with emotions and fear. Arthritis Rheum. 2007;56(4):1345-1354.

113.

Tagliazucchi

Balenzuela

Fraiman

Chialvo

. Brain resting state is disrupted in chronic back pain patients. Neurosci Lett. 2010;485(1):26-31.

114.

Fuchs

Kastner

Wagner

Hawes

Ebersole

. A standardized boundary element method volume conductor model. Clin Neurophysiol. 2002;113(5):702-712.

115.

Lancaster

Tordesillas-Gutiérrez

Martinez

, et al. Bias between MNI and Talairach coordinates analyzed using the ICBM-152 brain template. Hum Brain Mapp. 2007;28(11):1194-1205.

116.

De Ridder

Vanneste

Smith

Adhia

. Pain and the triple network model. Review. Front Neurol. 2022;13.

117.

De Ridder

Verplaetse

Vanneste

. The predictive brain and the “free will” illusion. Front Psychol. 2013;4(131).

118.

Collura

. BrainAvatar is the next generation of BrainMaster software, supported by the Atlantis and Discovery EEG devices. It is the successor to Atlantis. 3(7), i, and Discovery 1.5. 9, and is designated the 4.0 release. It includes all existing features, and can run all existing protocols, including peripheral biofeedback, but include several innovative new.

119.

Sherlin

Arns

Lubar

, et al. Neurofeedback and basic learning theory: implications for research and practice. J Neurother. 2011;15(4):292-304.

120.

Arns

Batail

Bioulac

, et al. Neurofeedback: one of today's techniques in psychiatry? Encephale. 2017;43(2):135-145. doi:https://doi.org/10.1016/j.encep.2016.11.003

121.

Maheshkumar

Dilara

Maruthy

Sundareswaren

. Validation of PC-based sound card with biopac for digitalization of ECG recording in short-term HRV analysis. N Am J Med Sci. Jul 2016;8(7):307-311.

122.

Nurmiati

Saehana

Wahyono

. Development of simple teaching aid on resonance topic using audacity software as learning media. J Phys Conf Ser. 2021;2126(1):12005.

123.

Thompson

. An overview of audacity. General Music Today (Online). 2014;27(3):40-43.

124.

Delorme

Makeig

. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods. 2004;134(1):9-21.

125.

Delorme

Mullen

Kothe

, et al. EEGLAB, SIFT, NFT, BCILAB, and ERICA: new tools for advanced EEG processing. Comput Intell Neurosci. 2011;130714. https://doi.org/10.1155/2011/130714

126.

Bersagliere

Pascual-Marqui

Tarokh

Achermann

. Mapping slow waves by EEG topography and source localization: effects of sleep deprivation. Brain Topogr. 2018;31(2):257-269.

127.

Pascual-Marqui

. Discrete, 3D distributed, linear imaging methods of electric neuronal activity. Part 1: exact, zero error localization. 2007.

128.

Pascual-Marqui

Lehmann

Koukkou

, et al. Assessing interactions in the brain with exact low-resolution electromagnetic tomography. Philosop Trans of the R Soc A: Math, Phys EngSci. 2011;369(1952):3768-3784.

129.

Pascual-Marqui

. LORETA. University of Zurich; 2017.

130.

Gudmundsson

Runarsson

Sigurdsson

Eiriksdottir

Johnsen

. Reliability of quantitative EEG features. Clin Neurophysiol. 2007;118(10):2162-2171. doi:https://doi.org/10.1016/j.clinph.2007.06.018

131.

Vanneste

Song

De Ridder

. Thalamocortical dysrhythmia detected by machine learning. Nat Commun. 2018;9(1):1103.

132.

Han

Ridder

Vanneste

Chen

Y-C

Koo

J-W

Song

J-J

. Pre-treatment ongoing cortical oscillatory activity predicts improvement of tinnitus after partial peripheral reafferentation with hearing aids. Original research. Front Neurosci. 2020;14(410).

133.

Halder

Talwar

Jaiswal

Banerjee

. Quantitative evaluation in estimating sources underlying brain oscillations using current source density methods and beamformer approaches. eNeuro. 2019;6(4), ENEURO.0170-19.2019.

134.

Pascual-Marqui

. Standardized low-resolution brain electromagnetic tomography (sLORETA): technical details. Methods Find Exp Clin Pharmacol. 2002;24(Suppl D):5-12.

135.

Han S

Buchman

Arfanakis

Fleischman

Bennett

. Functional connectivity networks associated with chronic musculoskeletal pain in old age. Int J Geriatr Psychiatry. 2013;28(8):858-867.

136.

Hallett

de Haan

Deco

, et al. Human brain connectivity: clinical applications for clinical neurophysiology. Clin Neurophysiol. 2020;131(7):1621-1651.

137.

Krönke

K-M

Wolff

Shi

, et al. Functional connectivity in a triple-network saliency model is associated with real-life self-control. Neuropsychologia. 2020:107667. doi:https://doi.org/10.1016/j.neuropsychologia.2020.107667

138.

Modares-Haghighi

Boostani

Nami

Sanei

. Quantification of pain severity using EEG-based functional connectivity. Biomed Signal Process Control. 2021;69:102840. doi:https://doi.org/10.1016/j.bspc.2021.102840

139.

Cornell

. Accessing and analyzing data with Microsoft Excel. Microsoft Press; 2003.

140.

Motulsky

. Prism 4 statistics guide—statistical analyses for laboratory and clinical researchers. In Graphpad Software Inc. GraphPad; 2003:122-126.

141.

Sheskin

. Handbook of parametric and nonparametric statistical procedures. Chapman and Hall/CRC; 2003.

142.

Abdi

. Bonferroni and Šidák corrections for multiple comparisons. Encyclop Of Measure Stat. 2007;3:103-107.

143.

White

Balasubramaniam

Nayak

Barnett

. An observational analysis of the trope “A p-value of < 0.05 was considered statistically significant” and other cut-and-paste statistical methods. PLoS One. 2022;17(3):e0264360.

144.

Mitteer

Greer

Fisher

Cohrs

. Teaching behavior technicians to create publication-quality, single-case design graphs in graphpad prism 7: graphpad prism video tutorial. J Appl Behav Anal. 2018;51(4):998-1010.

145.

Hassan

Dufor

Merlet

Berrou

Wendling

. EEG source connectivity analysis: from dense array recordings to brain networks. PLoS One. 2014;9(8):e105041-e105041.

146.

Jatoi

Kamel

Malik

Faye

. EEG based brain source localization comparison of sLORETA and eLORETA. Australas Phys Eng Sci Med. 2014;37(4):713-721.

147.

Ong

W-Y

Stohler

Herr

. Role of the prefrontal cortex in pain processing. Mol Neurobiol. 2019;56(2):1137-1166.

148.

Seminowicz

Moayedi

. The dorsolateral prefrontal cortex in acute and chronic pain. J Pain. 2017;18(9):1027-1035.

149.

Nejati

Majdi

Salehinejad

Nitsche

. The role of dorsolateral and ventromedial prefrontal cortex in the processing of emotional dimensions. Sci Rep. 2021;11(1):1971.

150.

Rêgo

Lapenta

Marques

, et al. Hemispheric dorsolateral prefrontal cortex lateralization in the regulation of empathy for pain. Neurosci Lett. 2015;594:12-16.

151.

Lorenz

Minoshima

Casey

. Keeping pain out of mind: the role of the dorsolateral prefrontal cortex in pain modulation. Brain. 2003;126(5):1079-1091.

152.

Lieberman

Eisenberger

. The dorsal anterior cingulate cortex is selective for pain: results from large-scale reverse inference. Proc Natl Acad Sci USA. 2015;112(49):15250-15255.

153.

Russo

Sheth

. Deep brain stimulation of the dorsal anterior cingulate cortex for the treatment of chronic neuropathic pain. Neurosurg Focus. 2015;38(6):E11.

154.

Apkarian

Sosa

Sonty

, et al. Chronic back pain is associated with decreased prefrontal and thalamic gray matter density. J Neurosci. 2004;24(46):10410-10415.

155.

Davis

Moayedi

. Central mechanisms of pain revealed through functional and structural MRI. J Neuroimmune Pharmacol. 2013;8(3):518-534.

156.

Rodriguez-Raecke

Niemeier

Ihle

Ruether

May

. Brain gray matter decrease in chronic pain is the consequence and not the cause of pain. J Neurosci. 2009;29(44):13746-13750.

157.

Rodriguez-Raecke

Niemeier

Ihle

Ruether

May

. Structural brain changes in chronic pain reflect probably neither damage nor atrophy. PLoS One. 2013;8(2):e54475.

158.

Lim

Kim

Chung

. Increased low- and high-frequency oscillatory activity in the prefrontal cortex of fibromyalgia patients. Original research. Front Hum Neurosci. 2016;10(111).

159.

Emmert

Breimhorst

Bauermann

, et al. Active pain coping is associated with the response in real-time fMRI neurofeedback during pain. Brain Imaging Behav. 2017;11(3):712-721.

160.

Boccard

Fitzgerald

Pereira

, et al. Targeting the affective component of chronic pain: a case series of deep brain stimulation of the anterior cingulate cortex. Neurosurgery. 2014;74(6):628-637.

161.

Osaka

Morishita

Kondo

Fukuyama

. A word expressing affective pain activates the anterior cingulate cortex in the human brain: an fMRI study. Behav Brain Res. 2004;153(1):123-127.

162.

Maeoka

Matsuo

Hiyamizu

Morioka

Ando

. Influence of transcranial direct current stimulation of the dorsolateral prefrontal cortex on pain related emotions: a study using electroencephalographic power spectrum analysis. Neurosci Lett. 2012;512(1):12-16.

163.

Cannon

REX

Lubar

Congedo

Thornton

Towler

Hutchens

. The effects of neurofeedback training in the cognitive division of the anterior cingulate gyrus. Int J Neurosci. 2007;117(3):337-357.

164.

Witte

Kober

Ninaus

Neuper

Wood

. Control beliefs can predict the ability to up-regulate sensorimotor rhythm during neurofeedback training. Front Hum Neurosci. 2013;7:478.

165.

Rogala

Jurewicz

Paluch

Kublik

Cetnarski

Wróbel

. The do's and don'ts of neurofeedback training: a review of the controlled studies using healthy adults. Review. Front Hum Neurosci. 2016;10(301).

166.

Baliki

Petre

Torbey

, et al. Corticostriatal functional connectivity predicts transition to chronic back pain. Nat Neurosci. 2012;15(8):1117-1119.

167.

Bingel

Lorenz

Schoell

Weiller

Büchel

. Mechanisms of placebo analgesia: RACC recruitment of a subcortical antinociceptive network. Pain. 2006;120(1):8-15.

168.

Apkarian

Hashmi

Baliki

. Pain and the brain: Specificity and plasticity of the brain in clinical chronic pain. Pain. 2011;152(3 Suppl):S49-S64.

169.

Ohara

Crone

Weiss

Lenz

. Analysis of synchrony demonstrates ‘pain networks’ defined by rapidly switching, task-specific, functional connectivity between pain-related cortical structures. Pain. 2006;123(3):244-253.

170.

Napadow

LaCount

Park

As-Sanie

Clauw

Harris

. Intrinsic brain connectivity in fibromyalgia is associated with chronic pain intensity. Arthritis Rheum. 2010;62(8):2545-2555.

171.

Perneger

. What's wrong with Bonferroni adjustments. Br Med J. 1998;316(7139):1236-1238.

172.

Xia

Mingrui

Wang

Jinhui

Yong

Csermely

Peter

. BrainNet Viewer: A Network Visualization Tool for Human Brain Connectomics. PLoS ONE. 2013;8(7):e68910.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB

0.42 MB

0.53 MB