Systems, Man & Cybernetics - October 2016 - 13

to an interfaced device or allow the user to learn and
enhance finer control and coordination of one's limb
movement. These can have a significant impact on
clinical applications, (neurorehabilitation and neuroprostheses) as well as navigation applications (robot,
virtual reality, and games).
Notwithstanding the aforementioned highlights of this
area of research, it is worth mentioning its challenges and
limitations as well. SMR-BCI investigates different aspects
of voluntary movement, namely, preparation, intention,
execution, and imagination. The research on finer parameters of movement widely uses movement execution experiments. Even though studies on kinematics of movement
intention/imagination can have a significant role in rehabilitation applications, this area is yet to be explored. The
major challenges in this area include
1) BCI performance of imagined tasks; the SMR-BCI
research reports reduction in decoding accuracy for
imagined movement compared to executed movement,
owing to globalized activity
2) experiment paradigms that can elicit imagination of
finer movements; cues and feedback to ensure that subject performs kinesthetic imagination of such movement
will be challenging.
Some studies cited in this review [38], [39], [49], [87],
[100] report promising results in decoding imagined movement kinematics, and future research can look into this
aspect of SMR-BCI.
The performance of SMR-BCI is often reported in
terms of average classification accuracy or a correlation
coefficient over a limited number of subjects. However, to
ensure generalization of the results, analysis of statistical
significance maybe incorporated. Future research may
look into acquiring a larger dataset on standardized
experiment paradigms to study movement kinematics.
The report in [84] raised various concerns regarding the
use of linear regressors and a correlation coefficient as
the evaluation metric in movement kinematics reconstruction studies. The statistical analyses that can be
performed to avoid certain misinterpretations of the
results were further suggested. The study pointed out
how the usage of a linear regressor can limit the spectral
region of the brain signal under investigation and how the
correlation metric provided overly optimistic results in
lower spectral regions.
The studies in [85] and [86] investigated how the artifacts affect movement parameter decoding in EEG-BCI
systems. The effect of a slow trend on EEG signal and
proposed adaptive filtering methods to extract the same
was examined in [85]. In [86], the effect of eye movement
artifacts in the performance of linear decoding models
was demonstrated. The article reported significant fall
in performance of linear decoders as compared to nonlinear decoders once the electrooculography-related
activity was removed. The previously mentioned studies
point out the major design considerations in decoding

and signal enhancement algorithms and the need to
develop efficient alternatives.
Interesting research in this area is the study in [83]
that investigated the impact of decoding error in device
control using simulations and suggested how remapping
decoded parameters to control some other aspects of
device movement is a viable option. The study specifically reported that, for the same amount of error for position, velocity, and goal decoding, only the latter two
could produce accurate control output. It reported the
option of remapping the decoded velocity/goal to control
position of the output device or vice versa. As pointed
out by the authors, these interesting transformations
exist in our everyday skills, e.g., the car velocity being
controlled by foot position in the pedal.
Researchers still need to look into the possibility of
developing closed loop BCIs, which can provide real-time
high-dimensional motor control. Also, decoding of imagined or intended movement trajectory can also be explored
through design of proper experiment paradigms and neural
feature identification. It will also be interesting to see how
the various areas of a single limb contribute to gross movement and its parameters, which essentially combines the
goals of the various topics discussed in this review. Also,
incorporating a multimodal brain data acquisition system
(EEG-fNIRS, EEG-MEG) can enhance the signal resolution
and hence provide thorough insights into neurophysiological phenomena involved in these movement tasks.
Present-day BCI is capable of interpreting the electrophysiological or hemodynamic activity of the brain,
thereby establishing it as a possible augmentative communication and control technology for disabled people [1]-[4].
However, the major challenges in this research still include
unreliability of the BCI performance due to low signal resolution and nonstationary neural activity. The research
reviewed in this article suggest that translation of motor
intentions into precise control commands is possible with
the help of efficient BCI algorithms. Further studies [80],
[90] suggest the use of low-cost and user-convenient
commercial EEG amplifiers in SMR-BCI. These results
encourage research to develop real-life, practical, and
user-friendly BCI technology.
We would like to point out that even though the
research reported in this article promises feasibility of a
higher-dimensional motor control system, applying SMRBCI as a rehabilitation technology for users who are
locked in may encounter several difficulties. The adaptation of BCI systems to account for neural features affected by neuromuscular impairments will be challenging.
The review in [91] reports EEG-BCI research in stroke
rehabilitation using imagined and executed motor tasks.
BCI-based rehabilitation for finer motor control can also
have significant impact in motor recovery of the user
and, hence, extensive research is required for the same.
In conclusion, the SMR-BCI research indicated that finer
and precise motor control to an interfaced device could
O c tob e r 2016

IEEE SyStEmS, man, & CybErnEtICS magazInE

Table of Contents for the Digital Edition of Systems, Man & Cybernetics - October 2016