Characterizing the stimulation interference in electroencephalographic signals during brain–computer interface–controlled functional electrical stimulation therapy. Issue 3 (18th September 2021)

• Record Type: Journal Article
• Authors: Jovanovic, Lazar I.; Popovic, Milos R.; Marquez‐Chin, Cesar
• Is Part Of: Artificial organs. Volume 46:Issue 3(2022)
• Page Start: 398
• Page End: 411
• Publication Date: 2021-09-18
• Subjects: brain -- computer interface -- electroencephalography -- functional electrical stimulation -- noise analysis; Artificial organs -- Periodicals; 617.956
• DOI: 10.1111/aor.14059
• Languages: English
• ISSNs: 0160-564X

Abstract

Abstract: Introduction: The integration of brain–computer interface (BCI) and functional electrical stimulation (FES) has brought about a new rehabilitation strategy: BCI‐controlled FES therapy or BCI‐FEST. During BCI‐FEST, the stimulation is triggered by the patient's brain activity, often monitored using electroencephalography (EEG). Several studies have demonstrated that BCI‐FEST can improve voluntary arm and hand function after an injury, but few studies have investigated the FES interference in EEG signals during BCI‐FEST. In this study, we evaluated the effectiveness of band‐pass filters, used to extract the BCI‐relevant EEG components, in simultaneously reducing stimulation interference. Methods: We used EEG data from eight participants recorded during BCI‐FEST. Additionally, we separately recorded the FES signal generated by the stimulator to estimate the spectral components of the FES interference, and extract the noise in time domain. Finally, we calculated signal‐to‐noise ratio (SNR) values before and after band‐pass filtering, for two types of movements practiced during BCI‐FEST: reaching and grasping. Results: The SNR values were greater after filtering across all participants for both movement types. For reaching movements, mean SNR values increased between 1.31 dB and 36.3 dB. Similarly, for grasping movements, mean SNR values increased between 2.82 dB and 40.16 dB, after filtering. Conclusions: Band‐pass filters, used to isolate EEG frequency bands for BCI application, were also effective in reducing stimulation interference. In addition, we provide a general algorithm that can be used in future studies to estimate the frequencies of FES interference as a function of the selected stimulation pulse frequency, F STIM, and the EEG sampling rate, F S . Abstract : The report suggests that the functional electrical stimulation (FES) interference in the electroencephalographic (EEG) signals recorded during brain–computer interface (BCI) controlled FES therapy is a function of the selected stimulation pulse frequency and the EEG sampling rate. After identifying the frequency components of the FES interference, we conducted a signal‐to‐noise ratio analysis and found that band‐pass filters, used for the purpose of isolating the BCI‐relevant EEG bands, were effective in reducing the stimulation interference.