Is Implicit Motor Imagery a Reliable Strategy for a Brain–Computer Interface?

Abstract

Explicit motor imagery (eMI) is a widely used brain-computer interface (BCI) paradigm, but not everybody can accomplish this task. Here, we propose a BCI based on implicit motor imagery (iMI). We compared classification accuracy between eMI and iMI of hands. Fifteen able-bodied people were asked to judge the laterality of hand images presented on a computer screen in a lateral or medial orientation. This judgment task is known to require mental rotation of a person's own hands, which in turn is thought to involve iMI. The subjects were also asked to perform eMI of the hands. Their electroencephalography was recorded. Linear classifiers were designed based on common spatial patterns. For discrimination between left hand and right hand, the classifier achieved maximum of 81 ± 8% accuracy for eMI and 83 ± 3% for iMI. These results show that iMI can be used to achieve similar classification accuracy as eMI. Additional classification was performed between iMI in medial and lateral orientations of a single hand; the classifier achieved 81 ± 7% for the left hand and 78 ± 7% for the right hand, which indicate distinctive spatial patterns of cortical activity for iMI of a single hand in different directions. These results suggest that a special BCI based on iMI may be constructed, for people who cannot perform explicit imagination, for rehabilitation of movement, or for treatment of bodily spatial neglect.

Unimanual Versus Bimanual Motor Imagery Classifiers for Assistive and Rehabilitative Brain Computer Interfaces

“…The bimanual MI task had parietal ERD more spread over the non-dominant hemisphere but the intensity was not significantly stronger than during unimanual MI tasks. Contrary to the results of our previous study on another group of healthy participants [43] we found no significant difference between ERD during MI of the left and the right hand. That might explain somewhat lower classification accuracy of L-R classifier in this study vs 83±3% in [43].…”

“…Contrary to the results of our previous study on another group of healthy participants [43] we found no significant difference between ERD during MI of the left and the right hand. That might explain somewhat lower classification accuracy of L-R classifier in this study vs 83±3% in [43]. On average 20% people cannot use MI BCI [44] which might reflect on differences in classification accuracy between healthy volunteers.…”

Subject inefficiency phenomenon of motor imagery brain-computer interface: Influence factors and potential solutions

“…Mental rotation of a body part involves motor imagery of that body part, and both processes activate sensorimotor areas [84]. Because subjects are not necessarily aware of movement imagination during mental rotation, Osuagwu et al referred to it as implicit motor imagery [85]. Implicit motor imagery is an automatic process that does not depend on the capacity of consciously imagined movements.…”

Convolutional neural network based features for motor imagery EEG signals classification in brain–computer interface system

“…MI has been used widely in BCI. By imagining or performing a muscle movement action, the power of mu (8)(9)(10)(11)(12)(13) and beta (16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31) rhythms in the sensorimotor cortex will decrease or increase. Researchers hypothesized that an appropriate pattern of this phenomenon can be used as a suitable feature/signature in the classification.…”

“…In [16], after some pre-processing method like band pass filtering (BPF) and independent component analysis (ICA) to remove artifacts, CSP was employed to design spatial filters for each class of signal and LDA was used as a classifier. Osuagwu et al [16] achieved a maximum accuracy of 81 ± 8% for explicit MI and 83 ± 3% for implicit MI method in discrimination between left and right hand. Some other studies suggested combining some methods to achieve better results.…”