EEG-controlled functional electrical stimulation rehabilitation for chronic stroke: system design and clinical application

Abstract

Stroke is one of the most serious diseases that threaten human life and health. It is a major cause of death and disability in the clinic. New strategies for motor rehabilitation after stroke are undergoing exploration. We aimed to develop a novel artificial neural rehabilitation system, which integrates brain--computer interface (BCI) and functional electrical stimulation (FES) technologies, for limb motor function recovery after stroke. We conducted clinical trials (including controlled trials) in 32 patients with chronic stroke. Patients were randomly divided into the BCI-FES group and the neuromuscular electrical stimulation (NMES) group. The changes in outcome measures during intervention were compared between groups, and the trends of ERD values based on EEG were analyzed for BCI-FES group. Results showed that the increase in Fugl Meyer Assessment of the Upper Extremity (FMA-UE) and Kendall Manual Muscle Testing (Kendall MMT) scores of the BCI-FES group was significantly higher than that in the sham group, which indicated the practicality and superiority of the BCI-FES system in clinical practice. The change in the laterality coefficient (LC) values based on μ-ERD (ΔLC_m-ERD) had high significant positive correlation with the change in FMA-UE(r= 0.6093, P=0.012), which provides theoretical basis for exploring novel objective evaluation methods.

Keywords

brain–computer interface / functional electrical stimulation / electroencephalogram / laterality coefficient / chronic stroke

1 Introduction

Stroke is the leading cause of disability and is a leading cause of death in the United States [1]. It occurs when blood supply to the brain is compromised, which causes brain cell death and brain functional deficits. The physical disability can substantially affect the quality of life. An estimated 7.0 million Americans over 20 years of age self-report having had a stroke, and approximately 0.8 million people experience a new or recurrent stroke each year. Approximately 85% of patients who suffer and survive a new or recurrent stroke in the United States each year require rehabilitation [2]. From 2015 to 2035, the total direct medical stroke-related costs are projected to more than double, i.e., from $36.7 billion to $94.3 billion [1]. Despite advances in acute stroke care, the estimated direct and indirect costs of stroke continue to escalate and are disproportionally associated with long-term care and rehabilitation [3]. Accordingly, inexpensive and effective rehabilitation methods and equipment are urgently necessary for patients with stroke, their families, and even the whole society.

Studies have shown that neuroplasticity is an important process, which underlies substantial gains in motor function after stroke [4,5], and the combination of task-specific training and general aerobic exercise is still the gold-standard treatment for post-stroke rehabilitation. Exercise and training, including both passive and active motion exercises, have long been used to restore motor function after stroke [6]. Active motion exercise (AME) is effective for maintaining and promoting health, physical fitness, and functional independence, especially in terms of endurance, muscular strength, flexibility, and balance [7−9]. Passive motion exercise (PME) may be an alternative mode of exercise for patients who cannot perform AME. Machine-based PME is beneficial in frail users [10], as well as in community-dwelling patients with chronic stroke and spastic paralysis [11].

Functional electrical stimulation (FES) and brain-computer interface (BCI) are mainly used to facilitate and restore normal movement in patients with motor dysfunction caused by stroke or spinal cord injury [12−17]. Both of the two approaches have gone beyond passive exercise mode [18,19].

A training BCI (tBCI) system, which combines peripheral control based on subjective intention and neurofeedback, can repeatedly strengthen the normal nerve conduction pathway from the brain to the muscle groups, increase the causality of information flow between brain regions, and effectively promote the gradual recovery of the defective neural reflex arc. A tBCI presents a form of neurofeedback in clinical rehabilitation. The feedback mainly includes visual, auditory, tactile, and proprioceptive sensations, and the feedback mode will be optimized according to the extent of brain activation [15].

Accurately predicting the recovery effect of patients using clinical evaluation alone is difficult [20]. Clinical evaluation of motor function in the days following a stroke can help predict subsequent recovery. Neurophysiological and neuroimaging methods may reflect the structure and function of the motor cortex; thus, these characteristics may contribute in predicting the rehabilitation level of patients with stroke and provide useful clinical information for patients’ personalized rehabilitation design.

The combination of BCI technology with FES fully realizes the advantages of both in neuromodulation and neuroplasticity to improve the motor function in the clinic. For a BCI-FES device, FES is triggered only when the BCI system detects a user’s intent to move, thus synchronizing the brain activity with the movement generated by muscle contraction. The active mode of BCI-FES can strengthen the neuroplasticity and the central–peripheral connection, thus achieving the effect of neural functional reconstruction. BCI-FES may be more effective than BCI or FES alone in the clinic based on the above therapeutic mechanism [19].

Recently, BCI-FES systems are capable of neurorehabilitation, especially in improving motor function for patients after stroke [21−29]. Such systems have been regarded as a promising method for stroke rehabilitation [22−24] and functional restoration in spinal cord injury [30−32] and tetraplegic [33−35] patients. Nevertheless, the therapeutic effect and mechanism of the novel methods which introduce the BCI technology are still unclear. Moreover, research on the therapeutic effect of long-term rehabilitation training applied with BCI-FES system is limited. Most studies on clinical rehabilitation effect for patients with stroke refer to clinical scales (as FMA, WMFT, and MBI) or behavioral characteristics [24,25]. An advantage of BCI-based rehabilitation system is its ability to collect user data, such as EEG, and reflect the brain functional state. Biasiucci et al. investigated the effects of neuroplasticity by analyzing changes in brain functional connectivity [22]. Exploring characteristics of neuroplasticity via neurophysiological information is conducive to instructing patient rehabilitation exercise and an accelerating recovery.

Therefore, we aimed to develop an artificial neural rehabilitation system and investigate the rehabilitation effects of BCI-FES on motor function of the upper limb in patients with stroke and to verify the effectiveness of this system in stroke rehabilitation through clinical trials. Given the lack of objective clinical rehabilitation assessment methods for patients with stroke and physicians, we aim to explore characteristic parameters based on EEG that objectively reflect recovery levels to assist physicians in optimizing rehabilitation training.

2 Materials and methods

2.1 System setup

BCI-FES system comprised an EEG amplifier (independent development) and an electrode cap (Neuroscan, USA) that receive brainwave information, two monitor screens (one displayed the principal computer interface for the operation of the researchers, and the other performed prompt and feedback messages), a mainframe computer to record and process brainwave activity, FES equipment, and principal computer interface. EEG data of 32 electrodes were recorded at a sampling frequency of 250 Hz. Electrodes were placed on positions F7, F3, Fz, F4, F8, FC3, FC1, FCz, FC2, FC4, C3, C1, Cz, C2, C4, CP3, CP1, CPz, CP2, CP4, P7, P3, Pz, P4, P8, TP7, TP8, O1, Oz, and O2 of the 10/20 system (reference: F7; ground: P3). FES stimulation parameters, such as frequency and intensity, were set in advance according to the rehabilitation physician. Principal computer interface consisted of case management module for data collection and storage, calibration module to establish a classifier based on EEG both during motor imagery and rest periods, control module to trigger FES equipment, and evaluation module to record scale assessments of patients by the therapist. As an artificial neural rehabilitation device, this system constructed a complete artificial neural pathway beyond the patient’s body and obtained synchronized activities of the brain and muscles to promote the recovery of dysfunctional reflex arc by strengthening the normal pathway from the brain to muscles repeatedly. The procedure of the clinical test with the BCI-FES system is shown in Fig. 1.

Upon artifact removal, rest and motor imagery EEG were split into segments by trial triggers and offline analyzed for prediction model generation. Traditionally, we used common spatial patterns (CSP) algorithm to extract features with spatial information of the whole brain signals. CSP extracts spatial distribution components of each category from multi-channel EEG data. These patterns use the diagonalization of the matrix to find a set of optimal spatial filters for projection, which maximize the difference between the variances of two types of signals [36]. In the present study, one type consisted of EEG signals during the motor imagery of the affected limb and the other was resting EEG. To calculate the spatial filters, we split data of each trial into overlapping (interval of 0.5 s) time segments of 2 s length. In addition, we established a classifier to predict whether patients were in the state of motor imagery or relaxation by using support vector machine (SVM) algorithm.

Considering the rehabilitation effect of the system, we focused on reducing the false positive rate (FPR), and plotted probability density curves of decision value with two populations using the offline prediction model generated above. We set the decision threshold to ensure that FPR was not more than 10% for the online process, that is, the area of the probability density curve (blue curve) in the relaxation state was not more than 10% (Fig. 2).

During the intervention session, we collected 2 s overlapping (interval of 1 s) EEG segments in real time. The acquired EEG signals were filtered with band-pass filter, CSP features were extracted, and the decision value was calculated through the generated classifier. The calculated decision value was compared with the predetermined threshold value. If it was higher than the threshold value, the patient’s motor cortex was considered to be activating. Meanwhile, the screen presents the result of “success” to patients and sends control commands to trigger FES whose parameters were already set.

2.2 Subjects and design

The institutional research ethics committee of Tianjin People’s Hospital approved this study. Participants were recruited from Tianjin People’s Hospital. All subjects were informed of the complete experimental procedure before participating in the clinical test. Written informed consent was obtained upon enrollment.

This clinical test was single blind. A total of 32 patients with stroke and hemiplegia were included in this trial, and they were randomly assigned to one of the two groups (the BCI-FES group or the NMES group). Outcome measures, such as Modified Ashworth Scale (MAS), Brunnstrom, and Mini-Mental State Examination (MMSE), were measured and recorded before the intervention period, and no differences were noted between the groups in these outcomes before clinical training. Fig. 3 shows the flow chart for patient recruitment.

The inclusion criteria were as follows: (1) age range from 30 to 75 years, (2) suffering from stroke for not more than 5 years (post-stroke duration≤5 years), (3) consciousness is clear, vital signs are stable (clear consciousness and stable vital signs), (4) MAS score of 0–2, and (5) Brunnstrom stages of 1–4. Excluded participants were those not suitable for electrical stimulation, including pregnant women, patients with a cardiac pacemaker, those with severe cardiopulmonary dysfunction, water and electrolyte disorders accompanied by shock, severe osteoporosis, with recent fractures and broken bone that does not heal, infection fever, which has not yet been controlled, and skin ulcers in the scalp and treatment areas.

Potential participants were also asked to complete the MMSE questionnaire. If the potential participant’s score was less than 27, the person was excluded.

2.3 Assessment tests

Participants were required to complete the Fugl Meyer Assessment of the Upper Extremity (FMA-UE), Kendall Manual Muscle Testing (Kendall MMT), and Modified Barthel Index (MBI) as clinical outcome assessments by the same blinded rehabilitation physician before and after the whole intervention program, which lasted 3 weeks.

FMA-UE aimed to assess movement control, muscle strength, and reflex activity of the upper extremity in patients with post-stroke hemiplegia. It comprises 33 items and has a maximum total score of 66. Here, we only assessed the motor part of FMA-UE. Kendall MMT determines a patient’s ability to voluntarily contract a particular muscle. It can identify specific impaired muscles or muscle groups to provide information for proper treatment. It measures muscle power in percentage terms, and its maximum total score is 100%. MBI is the world’s most commonly used scale to measure activities of daily living (ADL). The MBI scores the degree of independence of a client from any aid and has a maximum total score of 100. Through these three scales, we can obtain a comprehensive state information from the three aspects of motor function, strength, and self-care ability.

2.4 Training protocol

Subjects in both BCI-FES and NMES groups received intervention training four times per week for approximately 3 weeks (11 times training in total). For BCI-FES group, each treatment lasted approximately 40 min, which comprised calibration session and intervention session, excluding preparation and device setup. The calibration session is similar to training of motor imagery. During calibration session, patients were requested to perform two to four runs of 10 wrist-extension movement imagery trials, including 6 s of imagery task and 6 s of relaxation. When the patient was ready for a new run, therapists would press a key to restart. During intervention session, the patients were required to complete approximately five runs of FES intervention controlled by BCI. They were encouraged to do as many runs as possible. Each run comprised 10 trials. Patients were provided 10 s to perform the wrist-extension movement imagery according to the cue presented on the monitor. During this period, FES would not be triggered until activity in the motor cortex of the patient was identified. Once triggered, FES stimuli would last 6 s, followed by a 5 s rest period. The skin electrode patches were attached on the two ends of the extensor carpi ulnaris. The intensity of the FES stimuli was set to achieve wrist extension, and the stimulus frequency was set to 25 Hz. For NMES group, patients received two sessions of conventional low-frequency NMES treatment for a total of 40 min, during which 3−10 min of rest was provided. The electrode slices were attached to the skin above the two ends of extensor carpi ulnaris, and the stimulus frequency was set to 5 Hz. The rehabilitation physician determined stimulation current intensity of NMES based on the patient’s condition and tolerance to the electrical stimulator. Normally, it was between 35–45 mA.

EEG data were collected during the training period (calibration sessions and intervention sessions) for the BCI-FES group only. For NMES group, resting EEG was recorded both at the first and the last treatment.

2.5 Statistics analyses of outcome measures

We conducted statistical analysis of the outcome measures through MATLAB software. The clinical demographic data were analyzed by descriptive statistics to compare BCI-FES group with NMES group. Classification variables, such as sex and injury side, were analyzed using Fisher’s exact test; continuous variables, such as age, time since stroke, and other outcome measures, which were normally distributed, were analyzed using independent t-tests. Changes (pretest-posttest) in outcome measures, such as FMA-UE, Kendall MMT, and MBI, were calculated to test the differences between the two groups. Prior to this test, we compared outcome changes within each group using the paired t-test method. Results were accepted as statistically significant at P <0.05.

2.6 EEG data analysis

In this study, all EEG data sets from each session during the course of intervention were analyzed. During pre-processing, EEG was band-pass filtered using a 3th order noncausal Butterworth filter from 8 Hz to 30 Hz and downsampled to 200 Hz. Then, the processed data were filtered with the common average reference (CAR) for further analysis. To ensure the consistency of the data to be analyzed among individuals, especially considering the lesion site of the patients involved in this study, we referred to the method used in the study of Biasiucci et al. [22]. For all subjects, if his or her lesion was located in the right hemisphere, the EEG channels needed to flip horizontally on the spatial layout; thus, channels in the right hemisphere were localized over the left hemisphere. As such, electrode C3 was artificially positioned to remain in the lesional hemisphere for all subjects.

The event-related desynchronization (ERD) values of the main electrodes were calculated using event-related spectral perturbation (ERSP). The ERSP algorithm could inspect the changes in EEG power relative to the resting state from the perspective of time-frequency, thus extracting the ERD/ERS characteristics of MI. The ERSP method used in this study is as follows:

(1),\mathrm{ERSP}(f,t)={\displaystyle \frac{\mathrm{1}}{N}}{\displaystyle {\sum }_{i=\mathrm{1}}^N}|F_i(f,t){|}^{\mathrm{2}}

where N represents the number of trials, and F_i(f,t) is the spectral estimation of the ith trial at frequency f and time t [37].

For all electrophysiological analyses, we concentrated on the frequency bands m (8–13 Hz) and b (14–28 Hz), both of which are associated with sensorimotor rhythms of brain activities. Hence, according to Eq. (2), we measured the m and b ERD values by integrating the baseline-normalized ERSP values (dB) within the corresponding time intervals and frequency bands. The equation used is as follows:

(2),\mathrm{ERD}={\displaystyle \frac{\mathrm{1}}{n}}{\displaystyle {\sum }_{f\in F}}{\displaystyle {\sum }_{t\in T}}\mathrm{ERSP}(f,t)

where F represents the frequency band of ERD, T denotes the time interval ([0 6]s after the “Start” cue), and n is the number of time–frequency bins in a selected area.

Changes in ERD of both m and b bands (m-ERD and b-ERD, respectively) throughout therapy were analyzed. The quantitative analysis of the EEG data mainly involved LC and the topographic map based on EEG spectral power. Laterality coefficient (LC), which represented the hemispheric asymmetries of the ERD pattern, was calculated according to Eq. (3):

(3),LC=(C-I)/(C+I)

where C represents the ERD values of the hemisphere contralateral to the moved hand, and I denotes the values ipsilateral to the moved hand.

To track the variation tendency of motion-related cortical activity with the process of rehabilitation training, we conducted linear regression analysis on m-ERD and b-ERD at each time of training (11 times in total). In addition, we calculated the change (LC= LC_post− LC_pre) in LC values before and after the whole therapy (pre- and post-intervention) and explored the associations between LC changes and the changes in manual muscle, scores of MBI, and FMA-UE scale through Pearson correlation analysis.

3 Results

Before intervention, the two groups were statistically homogeneous. There was no significant difference in the primary outcomes such as the FMA-UE, MBI and Kendall MMT (independent t-tests, all P >0.05), and age, time since stroke (independent t-tests, all P >0.05), sex, injury side (Fisher’s exact tests, all P >0.05), and before training (Table 1) between the two groups.

Comparison of outcomes before and after intervention showed that FMA-UE, Kendall MMT, and MBI significantly increased (P <0.05) within each group. The results of between-group analyses are shown in Fig. 4. The increase in FMA-UE and Kendall MMT scores of the BCI group was significantly higher than that in the sham group (FMA-UE, independent t-tests, P= 0.009; MMT, independent t-tests, P=0.04). No significant difference was found for the MBI score (MBI, independent t-tests, P=0.8).

Scatter plots and lines of best fit of both m-ERD and b-ERD values at each time of intervention are shown in Fig. 5A and 5B. The m-ERD scatter plot represented high negative correlation (r-squared value= 0.6244, P=0.004), suggesting that the m-ERD values had a significant downward trend as the result of BCI-FES training. However, the b-ERD scatter plot showed lower negative correlation (r-squared value= 0.2694, P=0.025) compared with m-ERD values.

Finally, to test some fundamental assumptions of this study (that EEG-based features are able to represent post-stroke impairment of motor outputs), we compared LC values based on both m-ERD and b-ERD to outcome measures using Pearson’s r (Table 2). The change in LC values based on m-ERD (LC_m-ERD) showed a high positive statistically significant correlation with the change in FMA-UE (FMA-UE; r=0.6093, P=0.012) and MMT (MMT; r=0.61, P=0.011) scores. In addition, the change in LC values based on b-ERD (LC_b-ERD) exhibited a positive statistically significant correlation with FMA-UE (r=0.49, P=0.04).

4 Discussion

Three outcome measures (FMA-UE, MMT, and MBI) were employed for healing assessment. FMA-UE and MMT scores before and after 3 week-intervention within and between groups were significantly different. Nevertheless, no significant differences were found in outcome measure of MBI scores. As such, MBI was less sensitive to rehabilitation assessment than FMA-UE and MMT. The cause might be related to the duration of intervention. MBI was used to evaluate activities of daily living, for instance, unscrewing the cap of a bottle with both hands. The motor function of multiple limbs needs to be restored to normal, and the movement needs to be consistent and coordinated to achieve the above daily activities. Thus, achieving significant improvement over a short treatment period is difficult. On the contrary, FMA-UE and MMT, as the direct manifestation of motor function or ability, improvement can be observed early in clinical practice.

Clinically, the scales are the primary and widely used approach for the stroke rehabilitation assessment. Neurofeedback approaches, such as functional magnetic resonance imaging (fMRI) or EEG, rely on brain imaging patterns and have great potential in brain rehabilitation; they can effectively and objectively reflect the brain activity following a stroke [38−42].

In the present study, we extracted the time–frequency characteristics based on EEG to analyze motor cortex activities and pinpoint possible mechanisms underlying the changes in outcome measures throughout the whole intervention course. ERD and LC values in different frequency band ranges were calculated. The results showed that m-ERD values had a significant downward trend during the rehabilitation course. Mu (m) rhythm reflects electroencephalography oscillations from 8 Hz to 13 Hz, which is associated with the activity of the sensorimotor cortex. Mu suppression is considered as a biomarker of activity in the human mirror neuron system (MNS) [43−45]. Therefore, the downward trend in the m-ERD values suggested that the activity of sensorimotor cortex (SMC) in the lesional hemisphere performed stronger over time, to some extent. M1 activation on the ipsilesional side plays a crucial role in motor recovery [46]. This phenomenon was significantly correlated with increase of FMA-UE and MMT scores. A similar work by Andreas et al. [47] yielded similar results. Progressive laterality coefficient was used to predict the change in FMA in the clinic, and significant linear correlations were found. These results further indicated that LC_m-ERD could be employed as a physiologic indicator for post-stroke rehabilitation assessment. In addition, FMA-UE was suggested one of the most reliable clinical rehabilitation rating scales in the present study, which was consistent with previous conclusions [48,49].

4.1 Limitations

In this study, we designed an artificial neural system for stroke rehabilitation combining MI-BCI and FES technology. We also conducted clinical tests in patients with chronic stroke. The results indicated that the combination of MI-BCI and FES was more efficient and more effective than traditional therapies. However, the current rehabilitation system for clinical use is only the most basic version, with many modules to be optimized. Review highlighted the potential of BCIs in neural rehabilitation applications and the importance of enhancing BCI-FES integration [50]. Studying the mechanism of neural plasticity, neural coding and decoding technology, and neural feedback technology is essential in promoting the clinical application of BCI-FES in the future.

5 Conclusions

We developed an artificial neural rehabilitation system for patients with stroke and proposed a real-time method to set the classification decision threshold according to the probability density curve of the decision value in rehabilitation training. Thus, the system always has an accurate feedback mode of brain-controlled electrical stimulation. Through clinical trials, we demonstrated that this system could restore upper limb motor function in patients with stroke by rehabilitation training. In addition, we found a strong correlation between EEG characteristics (LC_m-ERD) and state improvements, such as upper limb motor function and muscle power. The overall findings in this work could provide further objective guidance and design for the clinical rehabilitation training of stroke and are expected to assist researchers in further exploring the principle of neurological rehabilitation.

References

[1] Benjamin EJ, Muntner P, Alonso A, Bittencourt MS, Callaway CW, Carson AP, Chamberlain AM, Chang AR, Cheng S, Das SR, Delling FN, Djousse L, Elkind MSV, Ferguson JF, Fornage M, Jordan LC, Khan SS, Kissela BM, Knutson KL, Kwan TW, Lackland DT, Lewis TT, Lichtman JH, Longenecker CT, Loop MS, Lutsey PL, Martin SS, Matsushita K, Moran AE, Mussolino ME, O’Flaherty M, Pandey A, Perak AM, Rosamond WD, Roth GA, Sampson UKA, Satou GM, Schroeder EB, Shah SH, Spartano NL, Stokes A, Tirschwell DL, Tsao CW, Turakhia MP, VanWagner LB, Wilkins JT, Wong SS, Virani SS. Heart disease and stroke statistics—2019 update: a report from the American Heart Association. Circulation 2019; 139(10): e56–e528

[2] Yang Q, Tong X, Schieb L, Vaughan A, Gillespie C, Wiltz JL, King SC, Odom E, Merritt R, Hong Y, George MG. Vital signs: recent trends in stroke death rates—United States, 2000–2015. MMWR Morb Mortal Wkly Rep 2017; 66(35): 933–939

[3] Benjamin EJ, Blaha MJ, Chiuve SE, Cushman M, Das SR, Deo R, de Ferranti SD, Floyd J, Fornage M, Gillespie C, Isasi CR, Jiménez MC, Jordan LC, Judd SE, Lackland D, Lichtman JH, Lisabeth L, Liu S, Longenecker CT, Mackey RH, Matsushita K, Mozaffarian D, Mussolino ME, Nasir K, Neumar RW, Palaniappan L, Pandey DK, Thiagarajan RR, Reeves MJ, Ritchey M, Rodriguez CJ, Roth GA, Rosamond WD, Sasson C, Towfighi A, Tsao CW, Turner MB, Virani SS, Voeks JH, Willey JZ, Wilkins JT, Wu JH, Alger HM, Wong SS, Muntner P. Heart disease and stroke statistics—2017 update: a report from the American Heart Association. Circulation 2017; 135(10): e146–e603

[4] Nudo RJ, Wise BM, SiFuentes F, Milliken GW. Neural substrates for the effects of rehabilitative training on motor recovery after ischemic infarct. Science 1996; 272(5269): 1791–1794

[5] Taub E, Uswatte G, Elbert T. New treatments in neurorehabili-tation founded on basic research. Nat Rev Neurosci 2002; 3(3): 228–236

[6] Dimyan MA, Cohen LG. Neuroplasticity in the context of motor rehabilitation after stroke. Nat Rev Neurol 2011; 7(2): 76–85

[7] Nelson ME, Rejeski WJ, Blair SN, Duncan PW, Judge JO, King AC, Macera CA, Castaneda-Sceppa C. Physical activity and public health in older adults: recommendation from the American College of Sports Medicine and the American Heart Association. Med Sci Sports Exerc 2007; 39(8): 1435–1445

[8] Takeshima N, Rogers NL, Rogers ME, Islam MM, Koizumi D, Lee S. Functional fitness gain varies in older adults depending on exercise mode. Med Sci Sports Exerc 2007; 39(11): 2036–2043

[9] Yu W, An C, Kang H. Effects of resistance exercise using Thera-band on balance of elderly adults: a randomized controlled trial. J Phys Ther Sci 2013; 25(11): 1471–1473

[10] Takahashi T, Koizumi D, Islam MM, Watanabe M, Narita M, Takeshima N. Effects of passive exercise machine-based training on day care service user frail elderly. Rigakuryoho Kagaku 2011; 26: 209–213

[11] Takahashi T, Islam MM, Koizumi D, Narita M, Takeshima N. The effects of low intensity exercises performed by community-dwelling chronic stroke patients on passive movement-type machines. Rigakuryoho Kagaku 2012; 27(5): 545–551

[12] Rushton DN. Functional electrical stimulation and rehabilitation—an hypothesis. Med Eng Phys 2003; 25(1): 75–78

[13] Granat MH, Ferguson AC, Andrews BJ, Delargy M. The role of functional electrical stimulation in the rehabilitation of patients with incomplete spinal cord injury—observed benefits during gait studies. Paraplegia 1993; 31(4): 207–215

[14] Wassermann EM. Changes in motor representation with recovery of motor function after stroke: combined electrophysiological and imaging studies. EEG Clin Neurophysiol 1995; 97(4): S26

[15] Hochberg LR, Serruya MD, Friehs GM, Mukand JA, Saleh M, Caplan AH, Branner A, Chen D, Penn RD, Donoghue JP. Neuronal ensemble control of prosthetic devices by a human with tetraplegia. Nature 2006; 442(7099): 164–171

[16] Silvoni S, Ramos-Murguialday A, Cavinato M, Volpato C, Cisotto G, Turolla A, Piccione F, Birbaumer N. Brain-computer interface in stroke: a review of progress. Clin EEG Neurosci 2011; 42(4): 245–252

[17] Cervera MA, Soekadar SR, Ushiba J, Millán JDR, Liu M, Birbaumer N, Garipelli G. Brain-computer interfaces for post-stroke motor rehabilitation: a meta-analysis. Ann Clin Transl Neurol 2018; 5(5): 651–663

[18] Qiu S, Yi W, Xu J, Qi H, Du J, Wang C, He F, Ming D. Event-related β EEG changes during active, passive movement and functional electrical stimulation of the lower limb. IEEE Trans Neural Syst Rehabil Eng 2016; 24(2): 283–290

[19] Young BM, Williams J, Prabhakaran V. BCI-FES: could a new rehabilitation device hold fresh promise for stroke patients? Expert Rev Med Devices 2014; 11(6): 537–539

[20] Stinear CM. Prediction of motor recovery after stroke: advances in biomarkers. Lancet Neurol 2017; 16(10): 826–836

[21] Looned R, Webb J, Xiao ZG, Menon C. Assisting drinking with an affordable BCI-controlled wearable robot and electrical stimulation: a preliminary investigation. J Neuroeng Rehabil 2014; 11(1): 51

[22] Biasiucci A, Leeb R, Iturrate I, Perdikis S, Al-Khodairy A, Corbet T, Schnider A, Schmidlin T, Zhang H, Bassolino M, Viceic D, Vuadens P, Guggisberg AG, Millán JDR. Brain-actuated functional electrical stimulation elicits lasting arm motor recovery after stroke. Nat Commun 2018; 9(1): 2421

[23] Do AH, Wang PT, King CE, Abiri A, Nenadic Z. Brain–computer interface controlled functional electrical stimulation system for ankle movement. J Neuroeng Rehabil 2011; 8(1): 49

[24] McCrimmon CM, King CE, Wang PT, Cramer SC, Nenadic Z, Do AH. Brain-controlled functional electrical stimulation therapy for gait rehabilitation after stroke: a safety study. J Neuroeng Rehabil 2015; 12(1): 57

[25] Zhang X, Elnady AM, Randhawa BK, Boyd LA, Menon C. Combining mental training and physical training with goal-oriented protocols in stroke rehabilitation: a feasibility case study. Front Hum Neurosci 2018; 12: 125

[26] Kim T, Kim S, Lee B. Effects of action observational training plus brain-computer interface-based functional electrical stimulation on paretic arm motor recovery in patient with stroke: a randomized controlled trial. Occup Ther Int 2016; 23(1): 39–47

[27] Chung E, Park SI, Jang YY, Lee BH. Effects of brain–computer interface-based functional electrical stimulation on balance and gait function in patients with stroke: preliminary results. J Phys Ther Sci 2015; 27(2): 513–516

[28] Chung E, Kim JH, Park DS, Lee BH. Effects of brain-computer interface-based functional electrical stimulation on brain activation in stroke patients: a pilot randomized controlled trial. J Phys Ther Sci 2015; 27(3): 559–562

[29] Jang YY, Kim TH, Lee BH. Effects of brain–computer interface-controlled functional electrical stimulation training on shoulder subluxation for patients with stroke: a randomized controlled trial. Occup Ther Int 2016; 23(2): 175–185

[30] Abduallatif NA, Elsherbini SG, Boshra BS, Yassine IA. Brain–computer interface controlled functional electrical stimulation system for paralyzed arm. 2016 8th Cairo International Biomedical Engineering Conference (CIBEC), Cairo, 2016. 48–51

[31] Bockbrader M, Annetta N, Friedenberg D, Schwemmer M, Skomrock N, Colachis S 4th, Zhang M, Bouton C, Rezai A, Sharma G, Mysiw WJ. Clinically significant gains in skillful grasp coordination by an individual with tetraplegia using an implanted brain–computer interface with forearm transcutaneous muscle stimulation. Arch Phys Med Rehabil 2019; 100(7): 1201–1217

[32] Likitlersuang J, Koh R, Gong X, Jovanovic L, Bolivar-Tellería I, Myers M, Zariffa J, Márquez-Chin C. EEG-controlled functional electrical stimulation therapy with automated grasp selection: a proof-of-concept study. Top Spinal Cord Inj Rehabil 2018; 24(3): 265–274

[33] Osuagwu BC, Wallace L, Fraser M, Vuckovic A. Rehabilitation of hand in subacute tetraplegic patients based on brain computer interface and functional electrical stimulation: a randomised pilot study. J Neural Eng 2016; 13(6): 065002

[34] Colachis SC 4th, Bockbrader MA, Zhang M, Friedenberg DA, Annetta NV, Schwemmer MA, Skomrock ND, Mysiw WJ, Rezai AR, Bresler HS, Sharma G. Dexterous control of seven functional hand movements using cortically-controlled transcutaneous muscle stimulation in a person with tetraplegia. Front Neurosci 2018; 12: 208

[35] Pfurtscheller G, Müller GR, Pfurtscheller J, Gerner HJ, Rupp R. ‘Thought’—control of functional electrical stimulation to restore hand grasp in a patient with tetraplegia. Neurosci Lett 2003; 351(1): 33–36

[36] Ramoser H, Müller-Gerking J, Pfurtscheller G. Optimal spatial filtering of single trial EEG during imagined hand movement. IEEE Trans Rehabil Eng 2000; 8(4): 441–446

[37] Makeig S. Auditory event-related dynamics of the EEG spectrum and effects of exposure to tones. Electroencephalogr Clin Neurophysiol 1993; 86(4): 283–293

[38] Lioi G, Fleury M, Butet S, Lécuyer A, Barillot C, Bonan I. Bimodal EEG-fMRI neurofeedback for stroke rehabilitation: a case report. Ann Phys Rehabil Med 2018; 61: e482–e483

[39] Wang T, Mantini D, Gillebert CR. The potential of real-time fMRI neurofeedback for stroke rehabilitation: a systematic review. Cortex 2018; 107: 148–165

[40] Perronnet L, Lécuyer A, Mano M, Bannier E, Lotte F, Clerc M, Barillot C. Unimodal versus bimodal EEG-fMRI neurofeedback of a motor imagery task. Front Hum Neurosci 2017; 11: 193

[41] Savelov AA, Shtark MB, Mel’nikov ME, Kozlova LI, Bezmaternykh DD, Verevkin EG, Petrovskii ED, Pokrovskii MA, Tsirkin GM, Rudych PD. Dynamics of fMRI and EEG parameters in a stroke patient assessed during a neurofeedback course focused on Brodmann Area 4 (M1). Bull Exp Biol Med 2019; 166(3): 394–398

[42] Bönstrup M, Schulz R, Cheng B, Feldheim J, Thomalla G, Hummel F, Gerloff C. P108. The effect of task effort on recovery-related brain activity following motor stroke assessed with fMRI and EEG. Clin Neurophysiol 2015; 126(8): e102

[43] Pfurtscheller G, Neuper C, Andrew C, Edlinger G. Foot and hand area mu rhythms. Int J Psychophysiol 1997; 26(1–3): 121–135

[44] Hobson HM, Bishop DVM. Mu suppression—a good measure of the human mirror neuron system? Cortex 2016; 82: 290–310

[45] Perry A, Stein L, Bentin S. Motor and attentional mechanisms involved in social interaction—evidence from mu and alpha EEG suppression. Neuroimage 2011; 58(3): 895–904

[46] Grefkes C, Fink GR. Connectivity-based approaches in stroke and recovery of function. Lancet Neurol 2014; 13(2): 206–216

[47] Ray AM, Figueiredo TDC, López-Larraz E, Birbaumer N, Ramos-Murguialday A. Brain oscillatory activity as a biomarker of motor recovery in chronic stroke. Hum Brain Mapp 2020; 41(5): 1296–1308

[48] Bushnell C, Bettger JP, Cockroft KM, Cramer SC, Edelen MO, Hanley D, Katzan IL, Mattke S, Nilsen DM, Piquado T, Skidmore ER, Wing K, Yenokyan G. Chronic stroke outcome measures for motor function intervention trials: expert panel recommendations. Circ Cardiovasc Qual Outcomes 2015; 8(6 Suppl 3): S163–S169

[49] Pandian S, Arya KN. Stroke-related motor outcome measures: do they quantify the neurophysiological aspects of upper extremity recovery? J Bodyw Mov Ther 2014; 18(3): 412–423

[50] Boord P, Barriskill A, Craig A, Nguyen H. Brain–computer interface-FES integration: towards a hands-free neuroprosthesis command system. Neuromodulation 2004; 7(4): 267–276

RIGHTS & PERMISSIONS

Higher Education Press