Motor imagery (Page et al., 2001), or mental practice, attracted considerable interest as a potential neurorehabilitation technique improving motor recovery following stroke (Jackson et al., 2001). According to the Guidelines for adult stroke rehabilitation and recovery (Winstein et al., 2016), mental practice may proof beneficial as an adjunct to upper extremity rehabilitation services (Winstein et al., 2016). Several studies suggest that motor imagery can trigger neuroplasticity in ipsilesional motor cortical areas despite severe paralysis after stroke (Grosse-Wentrup et al., 2011; Shih et al., 2012; Mokienko et al., 2013b; Soekadar et al., 2015).
The effect of motor imagery on motor function and neuroplasticity has been demonstrated in numerous neurophysiological studies in healthy subjects. Motor imagery has been shown to activate the primary motor cortex (M1) and brain structures involved in planning and control of voluntary movements (Shih et al., 2012; Mokienko et al., 2013a,b; Frolov et al., 2014). For example, it was shown that motor imagery of fist clenching reduces the excitation threshold of motor evoked potentials (MEP) elicited by transcranial magnetic stimulation (TMS) delivered to M1 (Mokienko et al., 2013b).
As motor imagery results in specific modulations of brain electroencephalographic (EEG) signals, e.g., sensorimotor rhythms (SMR) (Pfurtscheller and Aranibar, 1979), it can be used to voluntarily control an external device, e.g., a robot or exoskeleton using a brain-computer interface (BCI) (Nicolas-Alonso and Gomez-Gil, 2012). Such system allowing for voluntary control of an exoskeleton moving a paralyzed limb can be used as an assistive device restoring lost function (Maciejasz et al., 2014). Besides receiving visual feedback, the user receives haptic and kinesthetic feedback which is contingent upon the imagination of a specific movement.
Several BCI studies involving this type of haptic and kinesthetic feedback have demonstrated improvements in clinical parameters of post-stroke motor recovery (Ramos-Murguialday et al., 2013; Ang et al., 2014, 2015; Ono et al., 2014). The number of subjects with post-stroke upper extremity paresis included in these studies was, however, relatively low [from 12 (Ono et al., 2014) to 32 (Ramos-Murguialday et al., 2013) patients]. As BCI-driven external devices, a haptic knob (Ang et al., 2014), MIT-Manus (Ang et al., 2015), or a custom-made orthotic device (Ramos-Murguialday et al., 2013; Ono et al., 2014) were used. Furthermore, several other studies reported on using BCI-driven exoskeletons in patients with post-stroke hand paresis (Biryukova et al., 2016; Kotov et al., 2016; Mokienko et al., 2016), but these reports did not test for clinical efficacy and did not include a control group. While very promising, it still remains unclear whether BCI training is an effective tool to facilitate motor recovery after stroke or other lesions of the central nervous system (CNS) (Teo and Chew, 2014).
Here we report a randomized and controlled multicenter study investigating whether 10 sessions of BCI-controlled hand-exoskeleton active training after subacute and chronic stroke yields a better clinical outcome than 10 sessions in which hand-exoskeleton induced passive movements were not controlled by motor imagery-related modulations of brain activity. Besides assessing the effect of BCI training on clinical scores such as the ARAT and FMMA, we tested whether improvements in the upper extremity function correlates with the patient’s ability to generate motor imagery-related modulations of EEG activity.[…]