This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Research Protocols, is properly cited. The complete bibliographic information, a link to the original publication on http://www.researchprotocols.org, as well as this copyright and license information must be included.
Neuropathic pain is a debilitating secondary condition for many individuals with spinal cord injury. Spinal cord injury neuropathic pain often is poorly responsive to existing pharmacological and nonpharmacological treatments. A growing body of evidence supports the potential for brain-computer interface systems to reduce spinal cord injury neuropathic pain via electroencephalographic neurofeedback. However, further studies are needed to provide more definitive evidence regarding the effectiveness of this intervention.
The primary objective of this study is to evaluate the effectiveness of a multiday course of a brain-computer interface neuromodulative intervention in a gaming environment to provide pain relief for individuals with neuropathic pain following spinal cord injury.
We have developed a novel brain-computer interface-based neuromodulative intervention for spinal cord injury neuropathic pain. Our brain-computer interface neuromodulative treatment includes an interactive gaming interface, and a neuromodulation protocol targeted to suppress theta (4-8 Hz) and high beta (20-30 Hz) frequency powers, and enhance alpha (9-12 Hz) power. We will use a single-case experimental design with multiple baselines to examine the effectiveness of our self-developed brain-computer interface neuromodulative intervention for the treatment of spinal cord injury neuropathic pain. We will recruit 3 participants with spinal cord injury neuropathic pain. Each participant will be randomly allocated to a different baseline phase (ie, 7, 10, or 14 days), which will then be followed by 20 sessions of a 30-minute brain-computer interface neuromodulative intervention over a 4-week period. The visual analog scale assessing average pain intensity will serve as the primary outcome measure. We will also assess pain interference as a secondary outcome domain. Generalization measures will assess quality of life, sleep quality, and anxiety and depressive symptoms, as well as resting-state electroencephalography and thalamic γ-aminobutyric acid concentration.
This study was approved by the Human Research Committees of the University of New South Wales in July 2019 and the University of Technology Sydney in January 2020. We plan to begin the trial in October 2020 and expect to publish the results by the end of 2021.
This clinical trial using single-case experimental design methodology has been designed to evaluate the effectiveness of a novel brain-computer interface neuromodulative treatment for people with neuropathic pain after spinal cord injury. Single-case experimental designs are considered a viable alternative approach to randomized clinical trials to identify evidence-based practices in the field of technology-based health interventions when recruitment of large samples is not feasible.
Australian New Zealand Clinical Trials Registry (ANZCTR) ACTRN12620000556943; https://bit.ly/2RY1jRx
PRR1-10.2196/20979
Approximately 50% of individuals with spinal cord injury (SCI) report ongoing neuropathic pain at or below the level of injury [
Although many brain regions are involved in the experience of neuropathic pain, Gustin and colleagues have identified the key role of the thalamus in the development and maintenance of neuropathic pain following SCI. They have found that neuropathic pain after SCI is associated with altered thalamic volume [
The disruption in thalamocortical rhythms (thalamocortical dysrhythmia) can be detected by surface electroencephalography (EEG) [
There is accumulating evidence that thalamocortical dysrhythmia can be self-regulated by neuromodulative interventions [
Three single-arm trials have demonstrated that BCI-N interventions can reduce SCI neuropathic pain [
Current BCI-N interventions for both SCI neuropathic pain [
The primary objective is to evaluate the effectiveness of a multiday course of BCI-N intervention in a gaming environment to provide pain relief for individuals with SCI neuropathic pain. The secondary objective is to assess the intervention’s effectiveness on participants’ pain through pain interference. We will also determine whether the BCI-N intervention improves mood, sleep quality, quality of life, and well-being. Lastly, we will explore the neural mechanisms underlying the effect of a BCI-N intervention on SCI neuropathic pain. In particular, we will measure resting-state EEG and levels of thalamic GABA content pre- and postintervention.
This study will be conducted based on a single-case experimental design (SCED) with multiple baselines across participants. The SCED is a powerful and effective method that is increasingly used in clinical trial designs [
The SCED method is based on assessing the dependent variables (in this case, pain intensity and pain interference) repeatedly for each of the participants across phases. The design of this study will be AB + follow-ups, where A refers to the baseline phase, B is the intervention phase, and both are followed by a follow-up phase. In addition, we will conduct a further follow-up phase 3 months after completion of the intervention. We will conduct and report the SCED study in accordance with the Single-Case Reporting Guideline in Behavioural Interventions (SCRIBE) 2016 Statement [
We will recruit 3 individuals with complete thoracic SCI (American Spinal Injury Association Impairment Scale A) for this study. Participants need to meet the following inclusion criteria: (1) aged 18 to 80 years, (2) having persistent neuropathic pain for 6 months or longer, (3) having an average pain intensity of 2 or more (out of 10) in the past week on a visual analog scale (VAS; with 0 cm reflecting no pain to 10 cm reflecting the maximum pain imaginable), (4) being medically stable, and (5) demonstrating an ability to use the VAS. With regard to the neuroimaging component of the study, we will exclude individuals who have metal objects inside their body (eg, stents, metal clips, implants, and shrapnel).
We will randomly assign 3 participants to different baseline durations of the SCED using a simple randomization method [
Study procedure. Each participant (P1-3) will be randomly allocated to 1 of 3 baseline periods. Each baseline phase will be followed by 20 days of a brain-computer interface (BCI)-based neuromodulative intervention over a 4-week period, and there will be a 1-week follow-up period for all participants. A further 1-week follow-up will take place 3 months after completion of the intervention.
Each participant will receive 30-minute daily sessions of the BCI-N intervention for 20 days over a 4-week period in their home. Each session will involve two 15-minute BCI-N interventions divided by a 5-minute break, and each session will start and finish with measurement of the resting-state EEG levels. The BCI-N treatment incorporates an interactive gaming interface (ie, NeuroGame), and a neuromodulation protocol targeted to suppress theta and low alpha (4-8 Hz) and high beta (20-30 Hz) band powers and to enhance high alpha (9-12 Hz) band power (
Neurofeedback loop of the brain-computer interface-based neuromodulative intervention (Floating Jellyfish). EEG: electroencephalogram.
The VAS will serve as the primary outcome measure of SCI neuropathic pain. We will ask the study participants to rate the average intensity of pain during 3 specific epochs each day, using a VAS. The VAS is a 10-cm horizontal line with “No Pain” at one end and “Maximum Pain Imaginable” at the other end. Respondents are asked to make a mark along the line that represents their pain intensity. At 12 o’clock (noon), participants will rate the average intensity of the pain they experienced from the time they woke up that day until noon. At 6 PM, they will rate their average pain intensity between noon and 6 PM. Finally, at the time they go to bed, they will rate the average of the pain intensity they experienced between 6 PM and the time they went to bed. The mathematical average of the 3 ratings will then be computed to represent that participant’s average daily pain intensity. If any ratings are missing, the score will be the average of the ratings obtained. Although consensus groups recommend the numeric rating scale over the VAS in pain clinical trials because some individuals have problems with understanding the VAS [
Pain interference will serve as a secondary outcome measure. We will assess the degree of pain interference by 6 items from the Brief Pain Inventory (BPI) [
We will have daily contact with the participants, both to address any questions they may have about the VAS and BPI measures, and to ensure they are completing their pain diaries and BPIs during the times they are supposed to.
We will collect generalization measures to evaluate whether the effect of the treatment extends beyond improvements in the primary and secondary outcomes [
We will administer the psychological questionnaires at 5 time points: (1) prior to the baseline phase, (2) on the last day of the baseline phase, (3) on the last day of the intervention phase, (4) on the last day of the first follow-up phase, and (5) on the last day of the second follow-up phase. We will collect the resting-state EEG and the neuroimaging data at 2 time points: prior to and following completion of the intervention.
Participants will complete the 36-item Short Form Health Survey modified for SCI (SF-36 walk-wheel) [
We will record resting-state EEG with the participants’ eyes closed (3 minutes) and eyes open (3 minutes) using the 24-channel EEG device SMARTING. The electrode placements are according to the standard 10-20 locations. The electrode impedance will be kept under 5 kΩ, and the sampling frequency will be 500 Hz.
Participants will lie supine, headfirst, on the bed of a 3-T magnetic resonance imaging machine (Ingenia; Philips) with their head immobilized in a 32-channel head coil. We will use multiplanar (axial, sagittal, coronal) reformats for voxel placement. GABA-edited Meshcher-Garwood Point Resolved Spectroscopy (MEGA-PRESS) [
We will collect participants’ feedback regarding perceived safety and feasibility of the intervention following each session of BCI-N treatment using study-specific questions (qualitative data). Following the completion of 20 sessions of BCI-N treatment, the participants will complete the Usefulness, Satisfaction, and Ease of Use [
We will analyze primary and secondary outcomes separately based on the SCED analysis. The SCED analysis mainly relies on visual inspection. However, we will inspect and analyze the outcome measures from this study using both visual analysis and supplementary statistical analysis [
The data across all phases will be scattered and visually analyzed using both within-phase and between-phase analyses [
One of the most robust nonoverlap methods is the Tau effect size [
Tau-U effect size [
We will calculate and extract total scores and subscores of the psychological questionnaires (SF-36 walk-wheel [
We will analyze resting-state EEG data using custom scripts based on the EEGLAB toolbox [
We will analyze the acquired spectra using the Java-based magnetic resonance user interface (jMRUI version 6.0; MRUI Consortium). First, we will remove the dominant water resonance using the Hankel Lanczos singular valve decomposition algorithm. The ON and OFF spectral subsets will be summed, producing single ON and OFF 68-ms subspectra for each spectra dataset. These 68-ms subspectra will then be subtracted, resulting in GABA-edited difference spectra to measure GABA concentration at 3.01 ppm. We will quantify GABA using AMARES, a nonlinear least-squares fitting algorithm operating in the time domain. Peak fitting for GABA will be performed after manually defining the center frequency and line width of the GABA peak and modelling the GABA peak as a singlet. We will use Lorentzian curves to obtain the peak amplitude for this resonance. The OFF spectral subsets will be summed, producing single OFF 68-ms subspectra for each spectra dataset to measure creatine concentration at 3.02 ppm. We will then phase the single OFF 68-ms subspectra with respect to both the zero- and first-order phase. Spectral fitting in AMARES will be performed after manually defining the center frequency and line width of the creatine peak and modelling the creatine peak as a singlet. We will use Gaussian curves to obtain the peak amplitude for this resonance. Lastly, the GABA to creatine ratios will be calculated and extracted for further analysis (see Reliable Change Index below).
We will evaluate changes in generalization measures such as power spectral density, GABA to creatine ratios, and total scores and subscores of the psychological questionnaires using the Reliable Change Index (RCI) [
RCI is a ratio of the actual observed difference by the standard error of the difference (SEdiff): RCI = [Mpost – Mpre]/SEdiff, and SEdiff = SD√2(1 – r), where SD is the standard deviation of the measurement and r is the reliability coefficient of the measure.
This clinical trial has been approved by the University of New South Wales Human Research Ethics Committee (approval number: HC190411) and the University of Technology Sydney Human Ethics Committee (approval number: ETH19-4090). Additionally, this study is registered through the Australian and New Zealand Clinical Trials Registry (registration number: ACTRN12620000556943). We plan to commence the trial in October 2020 and expect to publish the results by the end of 2021.
Preliminary data in support of BCI-N treatment for SCI neuropathic pain have been reported [
The BCI-N system of this study will address a key limitation of previous EEG neurofeedback interventions, which have mostly relied on a single form of virtual interaction [
SCED is a powerful design to establish guidelines for evidence-based interventions [
Investigating our novel self-developed BCI-N intervention with a SCED and evaluating it with visual and statistical analyses will provide a rigorous methodology for this study. The high internal validity of a well-implemented SCED study allows for the results of the data analyses to draw reliable conclusions about the effectiveness of the intervention [
The SCED trial with multiple baselines will not be able to demonstrate effectiveness of the BCI-N treatment if 1 of the 3 participants drops out during the baseline or intervention phases. To address this limitation, we will perform a mock BCI-N treatment session for each participant prior to commencement of the trial in order to increase compliance and ensure a high comfort level of the EEG headset, a comprehensive understanding of the treatment protocol, assessment procedures, and performance of gameplay during the neurofeedback session.
brain-computer interface
BCI-based neuromodulative
Beck Depression Inventory
Brief Pain Inventory
electroencephalography
γ-aminobutyric acid
Meshcher-Garwood Point Resolved Spectroscopy
Neuropathic Pain Scale
Reliable Change Index
State Anxiety Inventory
single-case experimental design
spinal cord injury
Single-Case Reporting Guideline in Behavioural Interventions
36-item Short Form Health Survey
visual analog scale
SMG, NH-S, and TN-J designed the clinical trial and wrote the manuscript. SMG, NH-S, TN-J, C-TL, CATC, AKS, and T-TND developed the game interface and signal processing algorithms. All authors critically revised the manuscript and approved the final manuscript.
This work was supported by the Craig H. Neilsen Foundation, Encino, CA, USA; Rebecca L. Cooper Medical Research Foundation, Sydney, Australia; Cross-Faculty Collaboration Scheme University of Technology Sydney, Sydney, Australia; and SPHERE Frontiers Technology Clinical Academic Group, Sydney, Australia.
None declared.