Neurological rehabilitation of stroke patients via motor imaginary-based brain-computer interface technology＊＊☆

Hongyu Sun , Yang Xiang , Mingdao Yang

1Department of Computer Science and Technology, Tongji University, Shanghai 201804, China

2College of Information and Electrical Engineering, Shandong University of Science and Technology, Qingdao 266510, Shangdong Province,China

3Key Laboratory of Embedded System and Service Computing, Ministry of Education of the People’s Republic of China, Tongji University,Shanghai 201804, China

4Qingdao Economic & Technical Development Area First People’s Hospital, Qingdao 266555, Shangdong Province, China

lNTRODUCTlON

Stroke is one of the leading causes of long-term disability[1-2].Stroke-related motor deficits can be rehabilitated to a large degree compared with deficits resulting from other neurological disorders or traumatic brain injuries.The development of brain-computer interface (BCI) technology to restore function for individuals with severe motor disabilities has received increasing attention.Importantly, two methods of BCI systems have been shown to facilitate rehabilitation of individuals in cases where disease or trauma has abolished or severely impaired muscle control[3].BCI systems can substitute for the loss of normal neuromuscular outputs by enabling interactions with environmentsviabrain signals rather than muscles.A mechanical-hand orthosis, controlled by ongoing electroencephalogram (EEG)activity and based on synchronous BCI, has been introduced in Austria[4].Nevertheless,BCI technology remains complex and has only recently been included in experimental studies.BCI methods have been shown to rehabilitate motor function by inducing activity-dependent brain plasticity; they could help to rebuild brain plasticity by affecting motor learning[5].The most credible,evidence-based framework for creating an effective motor re-learning intervention following brain injury is activity-dependent central nervous system plasticity[6-8].Activity-dependent central nervous system plasticity is not limited to healthy nervous systems and can occur with trauma or disease; plasticity can include changes at synaptic, neuronal, and circuit levels[9-11].It is thought that repetitive movement practice influences activity-dependent central nervous system plasticity to restore more normal function.In contrast, BCI-based approaches use EEG signals to encourage and guide central nervous system plasticity for the improvement of motor function.Evidence for neurological rehabilitation as a result of motor imaginary training has been demonstrated in stroke rehabilitation[12-14].In the case of moderate to severe motor deficits, motor imagery represents an intriguing, new, “backdoor” approach to access the motor system and rehabilitation at all stages of stroke recovery[15].Motor imaginary training can independently improve motor performance and induce similar cortical plastic changes[16], providing an useful alternative when physical training is not possible.However, BCI technology in neurological rehabilitation has been utilized to control robotically assisted rehabilitation arms to help patients with rehabilitation training or to control other devices to facilitate the patient’s life.This has been based on the hypothesis that repeated activation and deactivation of brain signals restore motor functions and induce brain plasticity[5].The present study proposes a novel method to improve the effects of neurological rehabilitation in stroke patients by playing a real-time motor imaginary-based computer game.To assess the rehabilitation effects of motor imaginary-based BCI, 20 patients (12 males, 8 females;age range: 26-61 years; post-stroke duration of 4-15 months) with initial supra-tentorial stroke were enrolled in the present study.The patients were assessed by Berg Balance Scale and the Holden Walking Classification before and after treatment.

RESULTS

Quantitative analysis of subjects

From October 11 to November 7, 2010, a total of 28 patients were screened.Eight patients (28.6%) dropped out during training and 20 patients provided informed consent and were included in the final analysis.

Patient baseline characteristics(Table 1)

Table 1 Patient baseline characteristics at the beginning of training

Neurological rehabilitation of stroke patients via motor imaginary-based BCl technology

To test the effect of this approach, a BCI system was employed, which used EEG activity evoked by the patient’s intent and translated the motor intention into a control signal to operate a simulated man walking in a computer game environment.Figure 1 shows the BCI system framework; the aim was to gain 2D control of the walking direction of a man in a tomb using only the patient’s EEG.The game environment is shown in Figure 2; the simulated man was required to collect as many coins as possible, avoid monsters and traps, and safely escape the tomb as fast as possible.Left-hand imagery represented turning to the left, right-hand imagery represented turning to the right, and foot imaginary represented moving forward.Enhanced traditional post-stroke recovery could occur in the computer game environments by providing feedback to aid rehabilitation.Pre- and post-training scores for balance and functional outcome were compared between subjects (n=20) who completed the training according to the protocol.Analysis revealed statistically significant differences at the end of training for all outcome measures (Figure 3).There were significant differences in the following measurements:Berg Balance Scale (21.900±2.989,vs.37.700±3.404,P＜0.01) in subjects (n=20) who completed training according to pre- and post-training protocols, as well as walking ability according to Holden Walking Classification(2.510±0.512,vs.3.810±0.524,P＜0.01).

Figure 1 Framework of the brain-computer interface system.Brain signals are acquired by electrodes on the scalp or in the head and are processed to extract specific signal features that reflect patient intent.These features are translated into commands that operate a simulate man walking in a computer game environment.

Figure 2 Game environment.Three arrows represent directions corresponding to left, right, and forward, respectively.

Figure 3 Effect of balance training on Berg Balance Scale (A) and walking ability (B).aP＜0.01, vs.pre-training.Data are expressed as mean±SD (n=20).Paired t-test reveals statistically significant improvement for all outcome measures at post-training.

DlSCUSSlON

The present study was designed based on the hypothesis that repeated activation and deactivation of brain signals restore motor functions and induce brain plasticity.The effect of training on balance was measured, as well as walking and the overall ability to perform daily living activities.Through this small sampling, a significant benefit of BCI training therapy was observed.There were significant differences between pre- and post-training scores for Berg Balance(P＜0.01) and walking ability (P＜0.01) on Holden Walking Classification.A total of 20/28 patients completed the 20 training sessions (30 min/d, 5 days per week for 4 weeks) on the BCI training system,which suggested that balance training with visual feedback on the BCI system was well tolerated by the stroke patients.

The present study differed from previous studies,because BCI technology in neurological rehabilitation was used to control robotically assisted rehabilitation arms to help patients with rehabilitation training or to control other devices, thereby increasing the patient’s quality of life[17-19].In comparison, the present BCI-based approach used EEG signals to encourage and guide central nervous system plasticity to improve motor function.Previous clinical studies[20-21]compared the effects of physical or occupational therapy interventions in isolation with an approach that combined physical and motor imaginary practice.These studies consistently found greatest improvements in motor performance with interventions that combine physical and mental practice,followed by physical practice alone, and then by motor imaginary practice alone, which was superior to no practice at all.In another study, real exercise practice was supplemented with chronic post-stroke hemiparesis with 30 minutes of imagery practice (such as reaching towards a cup) twice a week, which was the same as physical practice; results demonstrated that 11 individuals in the chronic stroke phase exhibited significant improvements with mentally rehearsed activities of daily living compared with only physical practice[22].

Several limitations exist in the present study, such as small sample size, lack of comparison therapy without severe cognitive deficits and with some ability to walk.Because of the lack of comparison therapies, it is not possible to determine whether motor imaginary training was better than the typical clinical care.Future studies are needed to evaluate how these conclusions apply to patients with different lesion locations, extension, etiology,or even chronicity.

In summary, motor imaginary-based BCI training in a computer game environment may be an effective approach for neurological rehabilitation of stroke patients.This strategy could be used in combination with other rehabilitation strategies, such as providing bodyweight support to assist balance in acute post-stroke phase.

SUBJECTS AND METHODS

Design

Self-controlled experiment.

Time and setting

Experiments were performed at the Rehabilitation Center of Qingdao Economic & Technical Development Area First People’s Hospital in China, from November 11 to December 7, 2010.

Subjects

Twenty-eight patients with initial supra-tentorial stroke were recruited from the Rehabilitation Center of Qingdao Economic & Technical Development Area First People’s Hospital.A total of eight patients (28.57%) dropped out during training, and 20 patients (12 males, 8 females;age range: 26-61 years; post-stroke duration of 4-15 months) provided informed consent and were included in the study.

Patients with recurrent strokes, bilateral hemispheric,cerebellar or brainstem lesions, receptive aphasia,significant cognitive deficits, or depression were included.

Significant visual field deficits or hemi-neglect and orthopedic problems affecting participation were excluded.

All patients agreed to participate in this study and provided written informed consent prior to training.The protocol was approved by the Qingdao Ethics Committee,and was conducted in strict accordance with theAdministrative Regulations on Medical Institution,formulated by the State Council of the People’s Republic of China[23].

Methods

EEG recordings

Patients were asked to sit in an armchair with both hands relaxed; they were asked to look at a 17” computer monitor approximately 1 m in front of the subject and at eye level.The EEG signals (13 channels) were recorded by a 16-channel high-performance and high-accuracy biosignal amplifier and acquisition/processing system(g.tec, Graz, Austria) using the following channels located at positions of 10-20 international electrode-positioning standard: FC3, FCZ, FC4, C5, C3,C1, CZ, C2, C4, C6, CP3, CPZ, and CP4.Skin-electrode junction impedances were maintained at＜5 kΩ.Signals were digitized at a sampling frequency of 500 Hz and band-pass filtered between 8-30 Hz.Data collection incorporated three stages: (1) patient preparation, (2) training data collection, and (3) test data collection.The paradigm required the patient to control a cursor moving on the monitor by imagining the movements of his right hand, left hand, or foot for 5 seconds with a 3-second break between trials.For each patient, data were collected over two sessions with a 15-minute break in between.The first session was conducted without feedback, and 60 trials (20 trials for each class) obtained in this session were used for training and analysis.A total of 150 trials (50 trials for each class) in the next session served as test data to provide online feedbacks.The averaged time-frequency distributions of five channels during three different motor imaginary are shown in Figure 4.

Figure 4 Averaged time-frequency distributions of five channels during three different motor imaginary.Three lines represent three class imagination tasks; five columns represent the positions of five channels.The apparent event-related desynchronization (ERD) can be clearly seen in Cp4, C4 position on the first line, which is sustained until the end of the task (a total of five seconds).In C3, Cp3 position event-related synchronization (ERS)appear until the end of the task, demonstrating that the right hemisphere exhibits a sustained ERD and a sustained ERS in the left hemisphere using the left-hand imagination task.Similarly, in the right hand imagination task, C4 exhibits sustained ERS and Cp3 with sustained ERD.

Real-time feedback and BCI training task

The training process consisted of repetitive epochs of triggered movement imaginary trials.Each trial began with a full blank screen.The duration of each trial was 8 seconds.During the first 2 seconds, while the screen was blank, the subject was in a “relaxed” state.At 3 seconds, a visual cue representing the mental task to perform was displayed at the center of the monitor for 5 seconds (Figure 5).Depending on the presented symbol(left arrow, right arrow, down arrow), the patient was instructed to perform different tasks: imagining movement of the left hand, right hand, or foot.Three progress bars with red colors began to simultaneously increase from three different directions.The value of each bar was determined by the accumulated classification results, which was updated ten times per second (every 100 ms).For example, if the current classification result was “foot,” then the “down” bar increased one step and the values of the other two bars were retained.At second 8, a true or false mark indicated the final trial result by calculating the maximum value of the three progress bars, and the subject was asked to relax and wait for the next task.

Figure 5 Online feedback paradigm of the three-class motor imagery tasks.Three arrows represent tasks corresponding to left hand, right hand, and foot imagination, respectively.The progress bars provide real-time visual feedback.

The mental tasks represented by visual cue were randomly chosen to avoid adaptation.During training,feedback was presented to ensure compliance.Visual feedback was presented using “GOOD” or “ERROR.”

Real-time motor imaginary-based computer games

Finally, stroke-induced plegia patients were recruited to participate in neural training (play real-time motor imaginary-based computer games).Through the computer game environment, based on real-time motor imaginary, patients with stroke and complete hand paralysis learned to control μ-rhythm synchronization and desynchronization through motor imagery of the paralyzed hand.Harnessing cortical activity generated by such imageryviaa non-invasive BCI device enhanced and rehabilitated mobility and sensory faculties impaired by stroke.

Berg Balance Scale and Holden Walking Classification

The training effects were evaluated using primary quantitative outcome measures of balance-Berg Balance Scale[24]and Holden Walking Classification[25]before and 4 weeks after treatment.The Berg Balance Scale evaluates 14 sitting and standing activities, each on a graded 5-point scale (0–4).The maximum score was 56; higher scores indicated better balance: 0–20,wheelchair bound; 21–40, walking with assistance;41–56, independent.

Statistical analysis

Data were collected at the beginning of the study(pre-training) and at the end of 20 training sessions(30 min/d 5 days per week for 4 weeks) (post-training).To assess the effects of training, statistical analysis was performed using the pairedt-test (SPSS, Chicago, IL,USA).Data were presented as mean±SD following a normal distribution, and significance level was set to 0.01.

Author contributions:Hongyu Sun wrote the manuscript.Yang Xiang critically reviewed the manuscript.Mingdao Yang performed data analysis and statistics.

Conflicts of interest:None declared.

Funding:This study was supported by theNational Natural Science Foundation of China, No.60970062; and the Shanghai Pujiang Program, No.09PJ1410200.

Ethical approval:The experiments were approved by Ethics Committee of Qingdao Economic & Technical Development Area First People’s Hospital in China.

[1]Sims NR, Muyderman H.Mitochondria, oxidative metabolism and cell death in stroke.Biochim Biophys Acta.2010;1802(1):80-91.

[2]Donnan GA, Fisher M, Macleod M, et al.Stroke.Lancet.2008;371(9624):1612-1623.

[3]Janis JD, Jonathan RW.Brain-computer interfaces in neurological rehabilitation.Lancet Neurol.2008;7(11):1032-1043.

[4]Pfurtscheller G, Guger C, Müller G, et al.Brain oscillations control hand orthosis in a tetraplegic.Neurosci Lett.2000;292(3):211-214.

[5]Kübler A, Birbaumer N.Brain-computer interfaces and communication in paralysis: extinction of goal directed thinking in completely paralysed patients? Clin Neurophysiol.2008;119(11):2658-2666.

[6]Nudo RJ.Mechanisms for recovery of motor function following cortical damage.Curr Opin Neurobiol.2006;16(6):638-644.

[7]Wolpaw JR, Tennissen AM.Activity-dependent spinal cord plasticity in health and disease.Annu Rev Neurosci.2001;24:807-843.

[8]Ziemann U, Ilic TV, Pauli C, et al.Learning modi fi es subsequent induction of long-term potentiation-like and long-term depression-like plasticity in human motor cortex.J Neurosci.2004;24(7):1666-1672.

[9]Li S, Carmichael ST.Growth-associated gene and protein expression in the region of axonal sprouting in the aged brain after stroke.Neurobiol Dis.2006;23(2):362-373.

[10]Cramer SC, Shah R, Juranek J, et al.Activity in the peri-infarct rim in relation to recovery from stroke.Stroke.2006;37(1):111-115.

[11]Teasell R, Bayona NA, Bitensky J.Plasticity and reorganization of the brain post stroke.Top Stroke Rehabil.2005;12(3):11-26.

[12]Braun SM, Beurskens AJ, Borm PJ, et al.The effects of mental practice in stroke rehabilitation: a systematic review.Arch Phys Med Rehabil.2006;87(6):842-852.

[13]Sharma N, Pomeroy VM, Baron JC.Motor imagery: a backdoor to the motor system after stroke? Stroke.2006;37(7):1941-1952.

[14]Jackson PL, Lafleur MF, Malouin F, et al.Potential role of mental practice using motor imagery in neurologic rehabilitation.Arch Phys Med Rehabil.2001;82(8):1133-1141.

[15]Page SJ, Levine P, Leonard A.Mental practice in chronic stroke:Results of a randomized, placebo-controlled trial.Stroke.2007;38(4):1293-1297.

[16]Lotze M, Halsband U.Motor imagery.J Physiol Paris.2006;99(4-6):386-395.

[17]Buch E, Weber C, Cohen LG, et al.Think to Move: a Neuromagnetic Brain-Computer Interface (BCI) System for Chronic Stroke.Stroke.2008; 39(3):910-917.

[18]Abdullah MZ.Neurological rehabilitation of stroke patients by means of a robotically assisted brain controlled interface.Malays J Med Sci.2009;16(2):1-3.

[19]Lo AC, Guarino PD, Richards LG, et al.Robot-assisted therapy for long-term upper-limb impairment after stroke.N Engl J Med.2010;362(19):1772-1783.

[20]Page SJ, Levine P, Sisto SA, et al.Mental practice combined with physical practice for upper-limb motor deficit in subacute stroke.Phys Ther.2001;81(8):1455-1462.

[21]Malouin F, Richards CL, Doyon J, et al.Training mobility tasks after stroke with combined mental and physical practice: a feasibility study.Neurorehabil Neural Repair.2004;18(2):66-75.

[22]Page SJ, Levine P, Leonard A, et al.Modified constraintinduced therapy in chronic stroke: results of a single-blinded randomized controlled trial.Phys Ther.2008;88(3):333-340.

[23]State Council of the People's Republic of China.Administrative Regulations on Medical Institution.1994-09-01.

[24]Berg K, Maki BE, Williams JI, et al.Clinical and laboratory measures of postural balance in elderly population.Arch Phys Med Rehabil.1992;73(11):1073-1080.

[25]Holden MK, Gill KM, Magliozzi MR, et al.Gait assessment for the neurologically impaired: reliability and meaningfulness.Phys Ther.1984;64(1):35-40.