This study investigated the feasibility of using mobile EEG for a neurofeedback training at patients’ home. The Motor Imagery (MI) training for stroke rehabilitation showed promising results.
Stroke is a medical condition that can impact an individual’s life. One of the biggest impacts are on motor disabilities.
Effective rehabilitation methods are pivotal in aiding stroke patients. New methods may help regaining patients’ motor functions and improving their quality of life.
Methodology
Participants
The study engaged three chronic stroke patients. Each patient has diverse lesion types and levels of residual movement capacity.
The patients practiced every other day for four weeks. Their demographics and medical backgrounds provided a small yet varied sample. Spanning 4 weeks, they performed trainings every other day.
Motor Imagery Neurofeedback Protocol
Experimental procedure
Each 60-minute session entailed a 25-minute EEG setup and questionnaire phase, followed by a 30-minute Motor Imagery training.
The trainings consisted of three blocks, each lasting about 8 minutes with breaks. Participants engaged in right- and left-hand MI tasks. OpenVibe software provided visual cues ensuring task accuracy.
EEG captured data from 24 scalp sites using mobile EEG SMARTING. The initial block didn’t provide neurofeedback (NF), but its data calibrated the NF for the subsequent block.
The NF approach acknowledged that chronic stroke often skews sensorimotor activation patterns. Thus, the training aimed to recalibrate cortical lateralization, especially during MI of the affected limb.
Experimental task
A visually engaging 2D display showed the extent of contralateral event-related desynchronization (ERD) and its lateralization (See image bellow).
The ball’s exact placement on this display was deduced from classifying EEG band-power features. Offline EEG data analysis was additionally conducted.
The trial structure for motor imagery (MI) starts with a fixation-cross, which, after 3 seconds, gets accompanied by a three-shade blue graphic (see the video above). In the NF blocks, a white circle resembling a ball moves in response to classifier output magnitudes: horizontally (green arrow) and vertically (orange arrow). Following each trial, there’s a fixation dot and a 4.5 to 6-second intertrial gap. An illustrative event-related desynchronization (ERD) time course shows the ball’s position relation with MI-induced brain activity. Specifically, the ball’s horizontal movement mirrors the MI classification of the contralateral (blue) versus ipsilateral (red) hand, while the vertical shift compares contralateral baseline to MI.Outcome Measures
The heart of the study was to estimate the efficacy of the training. For this purpose rigorous motor assessments provided quantitative data on motor function improvement.
Secondary outcome measures added layers to the evaluation.
Secondary measures included MI-induced ERD, AM-induced ERD, fMRI BOLD activity, and the nuanced changes in the corticospinal tract (CST) white matter integrity.
Main Findings
The results, focusing on the experience and progress of each patient:
Participant P20
This patient exhibited a pronounced increase in EEG lateralization during the MI of their affected hand (shown in figure below).
This lateralization indicated a shift in brain activity beneficial for stroke rehabilitation. The data further revealed that increased EEG lateralization was mirrored by a heightened activity in fMRI scans.
This finding signified a deeper and more extensive brain engagement. The heightened neural activity also led to a rebalanced CST integrity and tangible improvements in motor abilities.
Participants P21 & P22
These participants, while showing progress, had a more nuanced journey (The results of P21 are shown in the figure below).
They both manifested non-significant increases in EEG lateralization when engaging in MI.
While there were evident neural changes, these did not translate to discernible improvements in their motor functions. This raises questions about the varied impact of MI EEG-based neurofeedback training on each patient.
Results highlighted that lesion type and residual movement capacity may influence neurofeedback efficiency.
Aggregated results for patients P20 (left) and P21 (right). (A) Shows an individual lesion map in green, superimposed on the Johns Hopkins University (JHU) White-Matter Tractography Atlas displaying the ipsilesional corticospinal tract (CST) in blue and its contralesional counterpart in red. The axial slices are based on specified coordinates in the Montreal Neurological Institute (MNI) space. (B) Offers a diffusion tensor imaging depiction of both ipsilesional (blue) and contralesional (red) CST fibers, as viewed sagittally. (C) Illustrates the event-related desynchronization (ERD) linked to motor imagery (MI) of the affected hand over the training’s duration. It presents the average relative power percentage during the initial and final three sessions, differentiating between the hemispheres. Two gray-shaded areas spotlight specific MI events, with two vertical lines capturing a time interval of relevance. (D) Portrays the topographical ERD changes prompted by the attempted movement (AM) of the affected hand, comparing scenarios before and after the MI training using high-density EEG. (E) Displays functional magnetic resonance imaging (fMRI) activity juxtaposed against averaged fractional anisotropy (FA) values, pre and post-MI training. It isolates specific brain regions and activities, referencing Brodmann areas for clarity. (F) Provides a look into the motivation scores, derived from distinct sessions, and includes regression lines specifically for the low-density EEG sessions.Patient Feedback
Direct feedback from the patients provided qualitative insights into the efficacy and reception of the training:
All three participants communicated a favourable response to the training program. They found value in the challenges it presented, pushing them to engage with their motor disabilities.
The absence of any reported adverse effects was a significant positive. This ensures that the training method is safe and patient friendly.
Their ability to perform the exercises effectively and consistently endorsed the convenience and feasibility of a home-based approach.
This is pivotal as it indicates that the process, can be implemented in diverse home settings. Important finding was that the EEG data quality or patient comfort weren’t compromised.
Study implications
The overarching aim of the study was two-fold to ascertain the feasibility of regular home-based MI neurofeedback training for chronic stroke patients and to understand its potential impact.
The overwhelmingly positive patient feedback, coupled with the good quality EEG data captured in home settings, establishes the study’s feasibility.
The changes in cortical activation, especially related to MI of the affected hand, were consistently noted across all participants. This uniform observation reinforces the potential efficacy of the training.
However, the variance in outcomes, especially between Participant P20 and Participants P21 & P22, suggests that individual factors play a significant role in determining the success of the training.
The study raises the idea that a more tailored, patient-specific approach may yield better results. This would include encompassing individual hardware, software, and training plans, might yield even.
Conclusion
This pioneering study makes a compelling case for frequent home-based Motor Imagery EEG-based neurofeedback training.
This kind of training proved to be feasible and potentially transformative approach for rehabilitating chronic stroke patients.
While preliminary evidence indicates its potential in driving behavioural, functional, and structural changes, the study acknowledges the need for more expansive research.
Only larger scale clinical trials can validate, refine, and possibly mainstream this promising approach for stroke rehabilitation.
Original publication source: https://pubmed.ncbi.nlm.nih.gov/28677413/
