Home used, patient self-managed, brain-computer interface for the management of central neuropathic pain post spinal cord injury: usability study

Abstract

Background: Central Neuropathic Pain (CNP) is a frequent chronic condition in people with spinal cord injury (SCI). Previously, we showed that using laboratory brain-computer interface (BCI) technology for neurofeedback (NFB) training, it was possible to reduce CNP in people with SCI. In this study, we show results of patient self-managed treatment in their homes with a BCI-NFB using a consumer EEG device.

Methods: Users: People with chronic SCI (17 M, 3 F, 50.6 ± 14.1 years old), and CNP ≥4 on a Visual Numerical Scale.

Location: Laboratory training (up to 4 sessions) followed by home self-managed NFB. User Activity: Upregulating the EEG alpha band power by 10% above a threshold and at the same time downregulating the theta and upper beta (20-30 Hz) band power by 10% at electrode location C4. Technology: A consumer grade multichannel EEG headset (Epoch, Emotiv, USA), a tablet computer and custom made NFB software.

Evaluation: EEG analysis, before and after NFB assessment, interviews and questionnaires.

Results: Effectiveness: Out of 20 initially assessed participants, 15 took part in the study. Participants used the system for 6.9 ± 5.5 (median 4) weeks. Twelve participants regulated their brainwaves in a frequency specific manner and were most successful upregulating the alpha band power. However they typically upregulated power around their individual alpha peak (7.6 ± 0.8 Hz) that was lower than in people without CNP. The reduction in pain experienced was statistically significant in 12 and clinically significant (greater than 30%) in 8 participants. Efficiency: The donning was between 5 and 15 min, and approximately 10-20% of EEG data recorded in the home environment was noise. Participants were mildly stressed when self-administering NFB at home (2.4 on a scale 1-10). User satisfaction: Nine participants who completed the final assessment reported a high level of satisfaction (QUESQ, 4.5 ± 0.8), naming effectiveness, ease of use and comfort as main priorities. The main factors influencing frequency of NFB training were: health related issues, free time and pain intensity.

Conclusion: Portable NFB is a feasible solution for home-based self-managed treatment of CNP. Compared to pharmacological treatments, NFB has less side effects and provides users with active control over pain.

Trial registration: GN15NE124 , Registered 9th June 2016.

Keywords: Brain computer Interface; Central neuropathic pain; Neurofeedback; Spinal cord injury; Usability.

Fig. 1

Usability framework (Rhui et al. 2018)

Fig. 2

Research protocol. N presents the number of participants involved in each phase

Fig. 3

a A member of research team demonstrating correct placement of the headset. Long arm EEG electrodes, marked with the arrows were placed over the central cortex. The electrode from which NFB was provided was placed posteriorly with respect to the imagined vertical line (dashed red line in the figure) aligned with participants’ ears. The electrode was placed between electrode location C2 and C4, exact location varies slightly depending on the head size. Image presented in user manual created for patients. b BCI NFB system consisting of EEG headset and tablet

Fig. 4

Hardware and software system structure

Fig. 5

NFB application software. User access a pain diary screen from the main screen. Prior to NFB training participants enter their pain level and then go to the baselines setting screen. This is followed by NFB training using GUI1 or GUI2. After completing NFB training users return to the pain diary to enter the post NFB level of pain and return to the main screen to exit the application. The parameters in the EEG setup screen were typically set at the hospital by researchers and were password protected

Fig. 6

The average relative changes of PSD during neurofeedback over all NF training sessions (mean ± STD) for each single participant. The horizontal dot lines mark Δ10% change in relative power with respect to the baseline recording. Positive values show increase and negative values show decrease with respect to the baseline power. Note that the NFB task was to increase the power of alpha for 10% or more and to decrease the power of theta and beta band for 10% or more. a Theta (4–8 Hz) in blue, and “individual” theta in orange. b alpha (9–12 Hz) in blue and individual alpha in orange colour. c higher beta (20–30 Hz) in blue and “individual” higher beta in orange. The results of participants 6, 8, 9, 13, and 15 are missing because they did not use the BCI NFB at home. Asterisks show statistically significant differences with respect to the baseline (p = 0.05)

Fig. 7

Power spectrum density during baseline (PreNFB, dashed line) and during NFB (solid line) over one session in three representative participants

Fig. 8

QUEST User priorities, in percentage. Number of participants N = 9

Fig. 9

Main themes from interviews with participants

Fig. 10

Experience with using BCI hardware (N = 9)