The role of virtual reality in improving motor performance as revealed by EEG: a randomized clinical trial

Participants

The present randomized clinical trial was conducted according to the CONSORT guidelines [36]. The trial was designed as a pilot, prospective, assessor blinded, parallel group study, and was performed at the IRCCS Centro Neurolesi “Bonino-Pulejo” (Messina, Italy). Eligible patients were enrolled between October 2015 and February 2016, according to the following criteria: (i) age ≥ 55 years; (ii) a first-ever ischemic supra-tentorial stroke (confirmed by magnetic resonance imaging -MRI) at least 6 months before their enrollment; (iii) an unilateral hemiparesis, with a Muscle Research Council -MRC- score ≤ 3 (a score of 3 indicates that muscle strength is reduced so that the joint can be moved only against gravity with the examiner’s resistance completely removed; 2 = muscle can move only if the resistance of gravity is removed; 1 = only a trace of movement; 0 = no movement observed) [37]; (iv) ability to follow verbal instructions, with a Mini-Mental State Examination >24; (v) a mild to moderate spasticity of muscles of hip, knee, and ankle (according to a Modified Ashworth Scale, MAS, ≤2) [38]; (vi) ability to perform manual gait training with or without external devices (Functional Ambulatory Categories 0-4). (vii) no severe bone or joint disease; and (viii) no history of concomitant neurodegenerative diseases or brain surgery.

We preferred to select patients with a first-ever ischemic supra-tentorial stroke, as this represents a better model of stroke lesion to perform EEG analysis. In fact, multiple vascular lesions could represent a significant limit for data interpretation, as they can generate variable signals that can interfere with signal recording. For instance, it has been reported that hemispheric powers differ clearly in single acute ischemic episodes, but correlate less well with the subtle, multifocal, or more gradual changes [39]. Moreover, single lesion model is more suitable to study interhemispheric balance. Last, patients with multiple vascular lesions may have different functional recovery. For the same reason, we limited the age of inclusion to >55, because beneath this age it is necessary to take into account other several, additional risk factors (including hemostatic, inflammatory and autoimmune factors, and cardioembolic sources, such as patent foramen ovale) that altogether increase the risk of multiple vascular lesions and may account for heterogeneity of sample [40].

Sample size

The sample size estimate was based on extrapolations from previous studies examining the effects of VR on gait in patients with stroke [41–44]. Accordingly, we used the effect size (0.9) of the primary composite endpoint for calculations. Power was set at 80%, alpha at 5%; we accounted for a dropout rate of 10%. Using a relatively conservative estimation, a total of 50 subjects (25 in each group) would be required to detect a difference of at least 20% in the primary outcome between the two treatment groups assuming non-inferiority with moderate correlations among covariates (R-squared = 0.5).

Study design

Thirty-five patients were assessed for eligibility according to the inclusion/exclusion criteria. Then, 24 patients were equally randomized into the RAGT + VR or RAGT-VR group with a 1:1 allocation ratio. For randomization, sealed envelopes were prepared in advance and marked on the inside with a + VR or -VR. The intervention period of both groups was identical, five sessions per week for eight consecutive weeks, 45-min for each session. We registered the EEG in the morning of the second day of the first week (T_PRE) and the last day of the last week of treatment (T_POST). Clinic and kinematic assessment was performed immediately before and after the Lokomat training. In addition, a physiotherapist supervised patients’ cooperation and participation to Lokomat training. The experimenters who analyzed the data and the therapists who performed the clinical tests were blind on patient allocation. Moreover, patients were not informed on the content of the VR adopted in the two groups.

RAGT intervention

Prior to the Lokomat training, all participants became familiarized with the Lokomat-Pro (RAGT + VR) or Lokomat-Nanos (RAGT-VR) (Hocoma AG; Volketswil, Switzerland), and had individually adapted their body weight support (BWS), leg guidance force (GF), and foot-lifting straps (that assisted ankle dorsiflexion for adequate toe clearance during the swing phase). Gait speed was set at a maximum of 1.8 m/s. Patients were asked to walk with their maximal effort. Then, patients were randomly assigned to RAGT + VR or RAGT-VR group. Both groups performed 40 Lokomat sessions (each lasting 40-45 min), five times a week, between 9^am and 11^am. The RAGT + VR group received a visual feedback showing a VR run game where the patient had to collect or avoid objects, to motivate him/her to walk actively (Fig. 1). Each avatar’s leg movement corresponded to that performed by the patient. To this end, we employed an impedance-based control strategy [45, 46], which allows variable deviations from a given leg trajectory, thus making the orthosis feel compliant (instead of making the orthosis stiff, i.e., allowing only negligible deviations from the set trajectory). The Lokomat is equipped with potentiometers and force transducers, capable of providing feedback about joint movement and joint moment production, respectively. In the impedance-based control strategy, the joint moment acting on the subject’s leg (which is generated by the force transducers of the orthoses) is proportional to the angular deviation of the performed movement from the preprogrammed movement. Therefore, the patient perceives the stiffness of the orthosis based on the setting of the mechanical impedance value (i.e., zero impedance makes no feedback experienceable, whereas maximal impedance results in maximal stiffness, equal to the position controller). The stiffness of the orthoses was set to the lowest point where subjects reported feeling the inertia of the Lokomat segments (about 30% of the maximal Lokomat impedance).

On the other hand, the RAGT-VR group was not provided an avatar but a smile, indicating the goodness of the leg movements. The biofeedback of the Lokomat gait orthosis is based on the interaction torques between the participant and the orthosis. To this end, hip and knee linear drives are equipped with force sensors that measure the human-machine interaction forces at the hip and knee joints for stance and swing phase, which are required to keep the participant on a predefined gait trajectory. The weighting functions of the gait cycle were defined for each part so that the resulting biofeedback values increased for therapeutically desirable movements, i.e. hip and knee flexion-extension force. Thus, biofeedback force values (representing the physical activity of the participants) were positive when the patient was actively participating and negative when passively participating in Lokomat training (or when muscles were inappropriately or involuntary activated) [36]. We obtained an average biofeedback value for each joint in the hemiparetic leg.

EEG recording and preprocessing

EEG was recorded using a Brain-Quick System (Micromed; Mogliano Veneto, Italy), from standard 19 electrodes headset according to the International 10-20 system (Fp1, Fp2, F7, F3, Fz, F4, F8, T3, C3, Cz, C4, T4, T5, P3, Pz, P4, T6, O1, O2, ground on the forehead), for 10 min while performing Lokomat training. To monitor eye movements, an electro-oculogram (EOG) with a bipolar montage (one pair of electrodes traced horizontal eye movements, a second pair the vertical ones) was also collected. EEG end EOG were sampled at 500 Hz, high pass filtered at 1 Hz using a zerophase FIR filter (order 7500) to minimize drifts, low pass filtered at 200 Hz (zerophase FIR filter order 36), and referenced to Cz [49]. An adaptive filter was applied to allow real-time filtering of signals recorded from EOG [50, 51].

Electrode impedance was kept below 5kΩ. During the entire EEG recording (as well as during the entire gait training), an experimenter checked for possible signs of drowsiness (e.g., abrupt worsening in gait performance, closed eyes, increase of proportion of theta and alpha activity in the eyes-open condition) [52], which were counted (given that monotonous gait pattern provided by RAGT may tend to induce sleepiness, thus decreasing arousal that negatively affects gait training progress). Patients were prohibited from drinking coffee, smoking, and change their bedtime during the three days prior EEG recording.

Infomax independent component analysis (ICA) was computed on the preprocessed EEG signal to decompose neural and artefactual sources [53–57]. In detail, ICA was computed two times. First, 500 ms-segmented EEG signals were removed if its probability distribution exceeded the average distribution by 5 ± SD. Then, ICA was computed to reject epochs based on the probability distribution of the IC projections. Thus, EEG segments were re-filtered (8-40 Hz) and a second ICA was computed a second time. The so-obtained IC were grouped into clusters using a k-means algorithm (based on the feature vector of dipole location, power spectra, and scalp map). The IC closest to the cluster centroid was remained for each subject, so to have equal contribution of each subject to the cluster-wise analysis.

EEG analysis

EEG analysis consisted of the computation of the power spectral density (PSD) (using Welch’s Method) and the time-frequency analysis to evaluate Event-related-spectral-perturbations (ERSPs) for each IC [58].

EEG was segmented into 1.4 s epochs (−700;700)ms with regard to the heel strike (HS) (i.e., the first moment the foot comes into contact with the floor) [59], thus obtaining 428 epochs. About that, the force-sensing resistor of the Lokomat device detected the movement onset of both lower limbs, which was synchronized with the EEG data. Epochs were rejected by using an automatic artifact rejection method (epochs with values of [−100;100]μV, ≥5SD of the mean kurtosis value, ≥5SD of the mean probability distribution, drifts of ≥50 μV/epoch and with a R² limit ≤0.3, spectra deviating from the mean by ±50 dB in the 0-2 Hz frequency window and by [−100;50]dB in the 20-100 Hz frequency window), visually inspection for artifacts, and if the power perturbation in the 20–40 Hz band deviated by +25 or -100 dB from the baseline at least for one IC [57, 60]. Rejection rate was 5%. This low rate is not surprising, given that it has been shown that scalp EEG recording during low-speed treadmill walking is not invalidated by excessive artefacts [61]. After this, the segmented data were time-warped and averaged together for all strides, so that initial affected-side heel strike, unaffected-side toe off, unaffected-side heel strike, affected-side toe off, and the subsequent affected-side heel strike occurred at the same times [49]. Spectrum analysis was carried using a standard fast Fourier transform (FFT) algorithm (Hanning-window, frequency resolution 0.7 Hz) within ϑ (4-7 Hz), μ (8–12 Hz), β (12–30 Hz), low-γ (Lγ) (31-45 Hz), and high-γ (Hγ) (46-70 Hz) bands [62, 63], and related to the phases of the gait cycle [55]. We opted to analyze these rhythms as it has been reported a different, specific role of each oscillation in sensory-motor pattern [27–35, 64]. For instance, there is evidence for a difference between the low and high α oscillations, which express action execution and observation, respectively [65, 66].

Single trial spectograms were computed and time-warped (thus aligning the time-points for right and left heel strike) over trials using a linear interpolation function to generate gait cycle ERSPs (i.e., epochs were based on the heel strike events, being the unaffected-side, the affected-side, and the next unaffected-side heel strike time-warped to 0, 50%, and 100% of the gait cycle, respectively). Relative changes in spectral power were obtained by averaging the difference between each single-trial log spectogram and baseline (the mean IC log spectrum over all gait cycles per training) [49]. To visualize significant ERSP changes, deviations from the average gait cycle log spectrum were computed with a bootstrap method [56, 57]. For statistical concern, bandwidth ERSP of each IC were averaged within each 10% of the gait cycle (10-point ERPS curve, frequency resolution 0.7 Hz) [67]. The average log spectrum for all movement cycles was subtracted from the log spectrogram for each movement cycle. We thus calculated the resulting PSD changes from this baseline (defined as the percentage decrement, event-related desynchronization –ERD- and increment, event-related synchronization –ERS- as a function of the percentage of the normalized gait cycle) [53] for each band and electrode-group of interest (with regard to the areas of activation of MNS reported in the literature [68], i.e., ipsi and contralesional frontal -Fp1/F7/F3, Fp2/F8/F4-, central -T3/C3, T4/C4- and parieto-occipital -T5/P3/O1, T6/P4/O2) [69–73].

Source localization

The source localization approach allows examining brain activities in various sources at different temporal phases of motor control. Because of high temporal resolution of the EEG signals, brain activities before and after the movement onset can be localized in order to distinguish cortical activities related to both motor planning (movement preparation) and motor execution (corticospinal pathway activation). The Estimation of Current Densities was carried by using Low Resolution Brain Electromagnetic Tomography (LORETA; free release of LORETA-KEY alpha-software) [63, 74–76]. The main components detected with the ICA (signal-to-noise ratio   >  1) were chosen for the source reconstruction. The distributed current density model (LORETA) with L1 norm method (based on the Montréal Neurological Institute (MNI) brain MRI) was then applied to the ICA data [77, 78]. The sources were constrained to the reconstructed layer of the folded cortex [79, 80].

Outcome measures

The primary endpoint, with respect to VR efficacy in post-stroke condition, was the proportion of patients achieving a 20% improvement in lower limb gait and balance at the end of the training, as measured by the Rivermead Mobility Index (RMI), the Tinetti Performance Oriented Mobility Assessment (POMA), and the gait cycle-related ERSPs. Indeed, a 20% improvement correspond to a significant minimal detectable change in RMI [81] and POMA [82]. According to previous work on assisted gait training in post-stroke patients, these changes are paralleled by EEG signal modification of at least 20% to be significant [83, 84].

As secondary outcomes, we considered the global MAS score derived from the muscles of hip, knee, and ankle, the Hamilton Rating Scale for Depression (HRS), the hip and knee flexion/extension force measured by the RAGT device, the extent to which a patient felt him/herself entrained in the VR training (reported on a visual analogue scale -VAS- ranging from zero -not at all- to ten -very much), and the mean of the episodes of drowsiness.

Statistical analysis

The normal distribution of the data was evaluated with the Kolmogorov-Smirnov test. Baseline data were compared between the two groups using a Student t-test for continuous variables if data were normally distributed, whereas a Mann-Whitney U test was used for non-normally distributed ordinal scale. Likewise, Wilcoxon test, Mann-Whitney U test, or t-test were used for within-group and between-group comparisons, depending on the types of data measurements.

The ERD/ERS changes for each frequency band were assessed by means of three-way ANOVA for repeated measures, employing the factor time (two levels: T_PRE and T_POST) and electrode-set (three levels: frontal, central, and parieto-occipital) as within-subject factors, and group (two levels: RAGT + VR and RAGT-VR) as between-subject factor. Based on the significance of the F-value, post-hoc paired-sample t-tests were carried out to assess the significance of interactions (Bonferroni correction). A p-value <0.05 was considered significant.

BWS and GF were included as covariates in the ANOVA analysis. In fact, it has been reported that RAGT training usually implies a steady progression of BWS and GF across the training program, so it is necessary to update and analyze these parameters in relation to the assessment of the outcomes [85]. In fact, both BWS and GF can influence spatiotemporal movement characteristics, thus affecting functional gait pattern. Indeed, finely tuning BWS and GF may somehow improve possible spatiotemporal gait asymmetries. On the other hand, missing the correction of these parameters augments inter-limb gait asymmetry for an extended duration in people with stroke [86]. Besides the factor electrode-set (which was employed in the ANOVA analysis to carry the spatio-temporal analysis of EEG signals at scalp level), we also added the factors lesion localization as covariate in the ANOVA analysis (according to the localization within left or right frontal, parietal, occipital, and temporal lobe). Such factor was added to augment inter-subject evaluation, as the sample was non-homogeneous for stroke localization, which can affect EEG signals beyond the overhead electrodes [87], also influencing both motor deficit degree and recovery [88].

ERSPs were computed in each frequency range for RAGT + VR and RAGT-VR using the average gait cycle log spectrum computed from the RAGT-VR as common baseline. The gait cycle was divided into in two stationary (S1, 10–30%, and S2, 60–80%) and two transition phases (T1, 30–60%, and T2, 80–10%). The stationary phases correspond to the midstance (10–30%), initial swing (60–70%), and midswing phases (70–90%), whereas the transition phases correspond to the terminal stance (30–50%), preswing (50–60%), terminal swing (90–100%), and loading response (0–10%) [89]. An ANOVA for repeated measures with the factors time (in relation with the gait cycle phases) (eight levels: two PRE and POST stationary and two PRE and POST transition phases), electrode-set (three levels: frontal, central, and parieto-occipital), and group (two levels: RAGT + VR and RAGT-VR) was computed for each frequency band. Multiple comparisons were corrected controlling for false discovery rate (p < 0.05) [90]. Sphericity assumption violations were Greenhouse-Geisser corrected.

Participant flow

We summarized in the CONSORT flow diagram (Fig. 2): the numbers of participants who were randomly assigned, who received the intended treatment, and who were analyzed for the primary outcome; the losses and exclusions during periods of recruitment, randomization, and follow-up. All treated patients completed the robotic training without reporting any side effect (Fig. 2). The main analysis focused on the consequences of the avatar observation on the RMI, POMA, and gait-related ERS/ERD, and involved all the patients who were randomly assigned to the two groups. Twelve patients were enrolled in each group (RAGT + VR: mean age 60 ± 4 years, 7 males and 5 females, disease duration 8 ± 2 months; RAGT-VR: mean age, 63 ± 6 years, 7 males and 5 females, disease duration 8 ± 2 months) (Table 1). There were no significant differences concerning any parameter between the two groups. Mean and standard deviation values of all outcome measures are reported in Table 1.

Clinical and kinematic data

Electrophysiological data

We found three main areas of brain activation that were more evident in the RAGT + VR group as compared to the RAGT-VR group across the gait cycle (Fig. 3a): (1) BA6 (Tailarach coordinates -x,y,z- 5, −1, 60); (2) BA7 (−14, −56, 53); and (3) BA17 (−20, −88, 3).

Concerning frontal activation (Fig. 3b; Table 3), RAGT + VR induced significant μ/β ERSP changes from baseline to S1, T1, and S2 phase of the gait cycle, as reflected by the significant interaction between gait phases and training for gait cycle related modulations concerning RAGT + VR group (Table 4). In particular, we observed a frontal ERD before the heel strike, followed by a frontal ERS during the heel strike (Fig. 4; Table 4). This changes also presented a significant difference in scalp projections between the trainings (Table 5; Fig. 5), with a greater magnitude in RAGT + VR group.

t × g × e	RAGT + VR						RAGT-VR
	t × e		TPRE-TPOST differences related to gait cycle phases				t × e		TPRE-TPOST differences related to gait cycle phases
			S1	T1	S2	T2			S1	T1	S2	T2
21, <0.001	28, <0.001	C	<0.001	<0.001	0.03	<0.001	8, 0.01	C	<0.001	<0.001	<0.001	<0.001
		F	<0.001	<0.001	<0.001	<0.001		F	<0.001	<0.001	<0.001	<0.001
		PO	0.004	0.005	<0.001	<0.001		PO	<0.001

Legend: t time, g group, e electrode, C central, F frontal, PO parieto-occipital; S1 first stationary phase; S2 second stationary phase; T1 first transition phases; T2 second transition phase

band	t × g × e	RAGT + VR				RAGT-VR
		t × e	C	F		t × e	C	F
μ	8.6, <0.001	77, <0.001	71, <0.001	50, <0.001		16, <0.001	3, 0.007	5, <0.001
β	15, <0.001	72, <0.001	87, <0.001	55, <0.001		39, <0.001	8, <0.001	9, <0.001
Lγ	9.2, <0.001	66, <0.001		53, <0.001		12, <0.001		25, <0.001
Hγ	28, <0.001	37, <0.001		12, <0.001
ϑ	8.9, <0.001	57, <0.001	51, <0.001			10, <0.001	18, <0.001
band	electrode	group	PRE-POST differences related to gait cycle phases
			S1	T1	S2	T2
μ	C	RAGT + VR	3.3, 0.01	−14, <0.001	3.4, 0.01
		RAGT-VR	2.4, 0.04	−6.3, 0.001	3.3, 0.01
	F	RAGT + VR		5, 0.002	4.5, 0.004
		RAGT-VR		3.6, 0.01	2.7, 0.03
	PO	RAGT + VR	-3.4, 0.01	−5, 0.002	−3.4, 0.01	−3.6, 0.01
		RAGT-VR	-3.3, 0.01	4.5, 0.002	−3.3, 0.01	−2.7, 0.03
β	C	RAGT + VR	-26, <0.001	6, 0.001	4.8, 0.003
		RAGT-VR	-2.9, 0.02	3.9, 0.007	2.9, 0.02
	F	RAGT + VR	-24, <0.001	8.3, <0.001	5.3, 0.002
		RAGT-VR	-3.5, 0.01	3.5, 0.02	3.5, 0.01
ϑ	C	RAGT + VR	5.1, 0.002
		RAGT-VR	3, 0.02
Hγ	F	RAGT + VR	-3.5, 0.01	−2.6, 0.03	−5.2, 0.002
		RAGT-VR
	PO	RAGT + VR	-2.8, 0.02	−4.4, 0.004	−7, <0.001	−4.5, 0.003
		RAGT-VR
Lγ	C	RAGT + VR		6, 0.001	5.1, 0.002
		RAGT-VR		3.5, 0.02	3.8, 0.008
	F	RAGT + VR		15, <0.001	13, <0.001
		RAGT-VR		4.9, 0.003	4.9, 0.003

Legend: t time, g group, e electrode, C central, F frontal, PO parieto-occipital; S1 first stationary phase; S2 second stationary phase; T1 first transition phases; T2 second transition phase

band	RAGT + VR			RAGT-VR
	C	F	PO	C		F		PO
μ -ERD	5 < 0.001		11, <0.001	6, <0.001				4, <0.001
β -ERD	10, <0.001		4, <0.001	6, <0.001		6, <0.001		3, <0.001
μ -ERS	5, <0.001	8, <0.001		13, <0.001		10, <0.001
β -ERS	4, <0.001	13, <0.001
ϑ -ERS	5, <0.001		4, <0.001	6, <0.001				4, <0.001
Lγ -ERS	7, <0.001	4, <0.001		6, <0.001		22, <0.001
Hγ -ERD		4, <0.001	4, <0.001

Legend: t time, g group, e electrode, C central, F frontal, PO parieto-occipital

μ/β changes were paralleled by a significant Lγ-ERS around the heel strike in both the groups and an Hγ-ERD from S1 to S2 (i.e., ending soon after the heel strike) only in RAGT + VR. Detailed statistical data of the main interactions and effects, and the temporal ERSP changes across the gait cycle are summarized in Tables 4 and 5 and Figs. 4 and 5.

Concerning central areas activation (Fig. 3b; Table 3), we found a significant μ/β ERD/ERS at S1, T1, and S2, a brief ϑ-ERS in the S1 phase, as revealed by the ANOVA and the post-hoc tests (Tables 4 and 5; Figs. 4 and 5) showing that spectral power in μ/β band was significantly reduced in T1 and increased in S1, and this was more evident in RAGT + VR than RAGT-VR.

Finally, parieto-occipital activation (Fig. 3b; Table 3) was characterized by a significant difference for μ and Hγ (ERD) between the groups during the entire gait cycle (Tables 4 and 5; Figs. 4 and 5). Post-hoc tests showed that scalp projections were significantly strengthened in RAGT + VR as compared to RAGT-VR.

Clinical-electrophysiological correlations

We observed a significant correlation between RMI and POMA score improvement and central β-ERS (r = 0.895, p = 0.001 and r = 0.570, p = 0.04, respectively) and frontal Hγ-ERD magnitude (r = −0.831, p = 0.003 and r = −0.615, p = 0.04, respectively). Moreover, the improvement in hip force was significantly correlated with frontal Hγ-ERD magnitude (r = −0.802, p = 0.004). Clinical-demographic characteristics (age, gender, disease duration, comorbidity) did not influence the primary outcomes.