Table 1.

Summary of non-invasive BCI-controlled FES studies used for rehabilitation of upper-limb motor function

Study	Population	BCI	FES	Intervention	Main results
Daly et al. [34]	1 stroke patient (F, 43 y)	1-channel EEG trigger detected by signal power change using a threshold method to classify rest vs. active states	1-channel FES applied to facilitate finger extension movements	Case study intervention: 9 session of 45 min delivered 3 times per week for 3 weeks	Participant’s ability to produce voluntary finger movements was improved after 9 sessions
Mukaino et al. [93]	1 stroke patient (M, 38 y)	Multi-channel EEG trigger using an LDA classifier detected signal power change of a multi-feature space to classify rest vs. active states	1-channel FES applied to facilitate finger extension movements	Case study crossover control design: (i) BCI-FES or (ii) FES was delivered for 60 min over the course of 2 weeks (10 days in total)	Clinical improvements and muscle tone changes were seen after BCI-FES as well as lateralization of cortical activations and affected corticomuscular coherence
Li et al. [74]	15 stroke patients (BCI-FES: n = 8, 5 M/3 F, 67.0 ± 5.0 y; control: n = 7, 6 M/1 F, 67.1 ± 6.0 y)	Multi-channel EEG trigger using an SVM classifier was used to detect rest vs. active states	1-channel FES applied to facilitate wrist extension movements	Randomized controlled intervention: (i) BCI-FES or (ii) FES training was delivered three times per week for 8 weeks	Improvements in motor function, activation of bilateral hemispheres, and altered activation of the sensorimotor cortexes was shown after BCI-FES intervention.
Kim et al. [66]	30 stroke patients (BCI-FES: n = 15, 6 M/9F; 59.1 ± 8.1 y; control: n = 15, 6 M/9F, 59.9 ± 9.8 y)	2-channel EEG trigger detected attention-related sensory motor rhythm using a threshold to classify rest vs. active states	1-channel FES was applied to stimulate wrist extensor muscles of the affected upper-limb	Randomized controlled intervention: (i) BCI-FES or (ii) conventional therapy (control) was delivered for 30 min per session over 4 weeks	Improvements in functional mobility and range of motion, suggesting improved motor function, was shown after BCI-FES intervention.
Marquez-Chin et al. [80]	1 stroke patient (M, 64 y)	1-channel EEG trigger detected by signal power change using a threshold to classify rest vs. active states	Multi-channel FES facilitated reaching movements: (i) forward reaching/retrieving (ii) reaching to the mouth, and (iii) lateral reaching	Case study intervention: 40 sessions of 90 min of BCI-FES were delivered 3 times per week	Improvements in clinical scores as well as the changes in arm function were shown after 40 sessions.
Osuagwu et al. [107]	12 SCI patients (BCI-FES: n = 7; FES: n = 5; 12 M, 51.7 ± 18.4 y)	Multi-channel EEG trigger using an LDA classifier detected signal power changes of a feature space to classify rest vs. active states	Multi-channel FES was applied to facilitate hand extension or flexion of both hands during active states	Randomized controlled intervention: (i) BCI-FES or (ii) FES were delivered 3–5 times weekly for 1 h (20 sessions in total)	BCI-FES therapy results in better neurological recovery and improvements in muscle strength compared to FES
Biasiucci et al. [17]	27 stroke patients (BCI-FES: n = 14, 6 M/8F, 56.4 ± 9.9 y; control: n = 13, 10 M/3F, 59.0 ± 12.4 y)	Multi-channel EEG trigger using a Gaussian classifier was used to discriminate rest vs. hand extension states	1-channel FES was applied to facilitate hand extension movements	Randomized controlled intervention: (i) BCI-FES or (ii) FES were delivered two times per week for a period of 5 weeks (10 sessions in total)	Improvements in motor function were accompanied by increase in functional connectivity between motor areas in the affected hemisphere after BCI-FES
Jovanovic et al. [58]	1 stroke patient (M, 57 y)	1-channel EEG trigger detected by signal power change using a threshold to classify rest vs. active states	Multi-channel FES facilitated functional movements: (i) hand opening/closing, and (ii) arm reaching/retrieving (varied between sessions)	Case study intervention: Two 40 one-hour BCI-FES sessions (80 sessions in total) were delivered with 3 sessions per week	Improvements in clinical scores and functional capacity were shown after completion of 80 therapy sessions

BCI Brain-computer interface, EEG Electroencephalography, FES Functional electrical stimulation, LDA Linear discriminant analysis, SCI Spinal cord injury, SVM Support vector machine