Think to move: a neuromagnetic brain-computer interface (BCI) system for chronic stroke

Abstract

Background and purpose: Stroke is a leading cause of long-term motor disability among adults. Present rehabilitative interventions are largely unsuccessful in improving the most severe cases of motor impairment, particularly in relation to hand function. Here we tested the hypothesis that patients experiencing hand plegia as a result of a single, unilateral subcortical, cortical or mixed stroke occurring at least 1 year previously, could be trained to operate a mechanical hand orthosis through a brain-computer interface (BCI).

Methods: Eight patients with chronic hand plegia resulting from stroke (residual finger extension function rated on the Medical Research Council scale=0/5) were recruited from the Stroke Neurorehabilitation Clinic, Human Cortical Physiology Section of the National Institute for Neurological Disorders and Stroke (NINDS) (n=5) and the Clinic of Neurology of the University of Tübingen (n=3). Diagnostic MRIs revealed single, unilateral subcortical, cortical or mixed lesions in all patients. A magnetoencephalography-based BCI system was used for this study. Patients participated in between 13 to 22 training sessions geared to volitionally modulate micro rhythm amplitude originating in sensorimotor areas of the cortex, which in turn raised or lowered a screen cursor in the direction of a target displayed on the screen through the BCI interface. Performance feedback was provided visually in real-time. Successful trials (in which the cursor made contact with the target) resulted in opening/closing of an orthosis attached to the paralyzed hand.

Results: Training resulted in successful BCI control in 6 of 8 patients. This control was associated with increased range and specificity of mu rhythm modulation as recorded from sensors overlying central ipsilesional (4 patients) or contralesional (2 patients) regions of the array. Clinical scales used to rate hand function showed no significant improvement after training.

Conclusions: These results suggest that volitional control of neuromagnetic activity features recorded over central scalp regions can be achieved with BCI training after stroke, and used to control grasping actions through a mechanical hand orthosis.

Figure 1

Trial description for BCI training. Whole-head MEG data (153 or 275-channels) was continuously recorded throughout each training block. At the initiation of each trial, 1 of 2 targets (top-right or bottom-right edge of screen) appeared on a projection screen positioned in front of the subject. Subsequently, a screen cursor would appear at the left edge of the screen, and begin moving toward the right edge at a fixed rate. A computer performed spectral analysis on epochs of data collected from a preselected subset of the sensor array (3 to 4 control sensors). The change in power estimated within a specific spectral band was transformed into the vertical position of the screen cursor feedback projected onto the screen. At the conclusion of the trial, if the subject was successful in deflecting the cursor upwards (net increase in spectral power over the trial period) or downwards (net decrease in spectral power over the trial period) to contact the target, 2 simultaneous reinforcement events occurred. The cursor and target on the visual feedback display changed colors from red to yellow. At the same time, the orthosis initiated a change in hand posture (opening or closing of hand). If the cursor did not successfully contact the target, no orthosis action was initiated.

Figure 2

Average group success rate as a function of training session. The average success rate for the last training session is 72.48 ± 18.36% (median ± interquartile range). As the total number of training sessions completed by patients was unique, the time-series for each individual was resampled and normalized to 20 sessions (the mode of the session duration across the patient group) using linear interpolation, before being averaged. The gray shaded area represents the 95% CI of the median estimate, which was computed using a bootstrap technique repeated 10 000 times. The boxplot (preand post-training median and interquartile range) inset shows a significant group increase in success rate between the first and last training sessions.

Figure 3

Individual subject performance, task-related brain activity, and lesion representations. Each row displays individual data for study participants. (From Left to Right): Column A shows the session performance for each patient. The gray shaded represents the 95% CI of the mean, which was computed using a bootstrap technique repeated 10 000 times. Columns B and C display task-related MEG brain activity from the sessions indicated by the red circle in Column A. With the exception of WF, whose performance peaked within the first 5 sessions of training, these represent the session with the highest performance that occurred within the final 4 sessions of training. Column B displays a flat map of the spectral amplitude difference across the MEG array between both target conditions. The sensor locations used to generate feedback and control the orthosis action are highlighted by the green-filled circles. The locations of central and parietal sensors within the right and left hemispheres of the arrays are outlined in white (labeled in the top row of Figure 3B as “C” and “P”, respectively). Column C displays a statistical map (R²) of the correlation of μ rhythm amplitude across the MEG array with target location/orthosis action. Column D displays single axial images from T1-weighted, high resolution MRI scans obtained for each subject (neurological convention). The red circles highlight the location of each patient’s lesion. All but patient GF had right hemisphere lesions.