| |
| |
| 4.
BCI - Brain–computer interfaces |
| |
4.2. Present-day BCIs
| |
4.2.4. Mu and beta rhythms
|
|
This BCI system is also based on ERD
and ERS of mu
and beta rhythms.
Research up to the present has focused on distinguishing between
the EEG associated with imagination of different simple
motor actions, such as right or left hand or foot movement,
and thereby enabling the user to control a cursor or an orthotic
device that opens and closes a paralyzed hand.
In the standard protocol, the user first participates in an
initial session to select a motor imagery paradigm. In each
of a series (e.g. 160) of 5.25 s trials, the user imagines one
of several actions (e.g. right or left hand or foot
movement, tongue movement) while EEG from electrodes
over sensorimotor cortex is submitted to frequency
analysis to derive signal features (e.g. the powers in the frequency
bands from 5 to 30 Hz).
For each imagined action, an n-dimensional feature
vector is defined. These vectors establish a user-specific linear
or non-linear classifier that determines from
the EEG which action the user is imagining. In subsequent sessions,
the system uses the classifier to translate the user’s
motor imagery into a continuous output (e.g.
extension of a lighted bar or cursor movement) or a discrete
output (e.g. selection of a letter or other symbol),
which is presented to the user as online feedback on a computer
screen.
Over 6–7 sessions with two-choice trials (i.e. left hand
vs. right hand imagery) users can reach accuracies over 90%.
About 90% of people can use this system successfully. The signal
features that reflect motor imagery and are used by the classifier
are concentrated in the mu- and beta-rhythm bands in EEG over
sensorimotor cortex.
Additional effort has been devoted to developing remote control
capabilities that allow the BCI to function in users’
homes while the classification algorithm is updated in the laboratory.
With this remote control system, a user paralyzed by a mid-cervical
spinal cord injury uses hand and foot motor imagery to control
an orthosis that provides hand grasp.
EEG over sensorimotor cortex is translated into hand opening
and closing by autoregressive parameter estimation and linear
discriminative classification.
Source:
Brain–computer
interfaces for communication and control, Clinical Neurophysiology
113 (2002) 767–791, Jonathan R. Wolpaw, Niels Birbaumer, Dennis
J. McFarland, Gert Pfurtscheller, Theresa M. Vaughan |
|
|
