LAUREYS 978-0-12-374168-4
Brain–Computer Interfaces for Communication
in Paralysed Patients and Implications for
Disorders of Consciousness
Andrea Kübler
O U T L I N E
Brain–Computer Interfaces: What, Why,
and Whereto 218
BCI for Communication and Control: Targeted
Patients 219
Brain–Computer Interfacing in Patients
With Motor Disability 221
Non-invasive BCI with the EEG as Input
Signal (EEG-BCI) 221
Invasive BCI 224
Non-Visual BCI 226
BCI in DOC 228
Learning in LIS and Non-responsive CLIS Patients 228
A Hierarchical Approach to Cognitive Processing
in Non-responsive Patients Including BCI 228
Problems and Prospects 230
Conclusion 231
Acknowledgements 231
References 231
C H A P T E R
ABSTRACT
Brain–computer interfaces (BCI) are direct connections between the brain and a computer. Regulation of
neuroelectrical activity or brain activity as a response to sensory stimulation are used to select items, words, or
letters in a communication programme or for neuroprosthesis control. Ten years work with severely paralysed and
locked-in patients demonstrated that BCI can be utilized for communication and interaction with the environment
if control of the motor periphery is lost. Recent non-visual BCI render this technology feasible for patients who
even lost control of eye movement due to injury or disease. In addition to passive stimulation and volitional
paradigms to assess cognitive processing in patients with disorders of consciousness (DOC), who may appear
quite similar to patients with motor paralysis, the use of BCI is suggested in this article. This review of BCI and
future prospects is a proposal to merge the so far independent streams of research – BCI in patients with paralysis
and cognitive processing in patients with DOC – for the benefit of the patients and to further elucidate how much
brain needs the mind.
p0110
p0120
217S. Laureys (Ed.) The Neurology of Consciousness: Cognitive Neuroscience and Neuropathology © 2009, Elsevier Ltd.
17
c0017
CH017.indd 217 5/10/2008 2:06:56 PM

218 17. BRAIN–COMPUTER INTERFACES FOR COMMUNICATION IN PARALYSED PATIENTS
III. COMA AND RELATED CONDITIONS
LAUREYS 978-0-12-374168-4 00017
BRAIN–COMPUTER INTERFACES:
WHAT, WHY, AND WHERETO
Brain–computer interfaces (BCI) allow us to interact
between the brain and artificial devices (for reviews
s0010
p0130
see for example [1, 24–26] ). They rely on continuous,
real-time interaction between living neuronal tissue
and artificial effectors ( Box 17.1 ). Neuronal activity of
few neurons or large cell assemblies is sampled and
processed in real time and converted into commands


p0050
p0060
p0070
p0080
A BCI system can be depicted as a series of functional
components [1, 2] . The starting point is the user, whose
intent is coded in the neural activity of his/her brain
(input). The end point is the device which is controlled
by the brain activity of the user (output) and provides
him or her with feedback of the current brain activity
(closed-loop systems).
Invasive recording methods allow us recording of: (1)
action potentials of single neurons with electrodes con-
taining neurotrophic factors inducing nerve growth into
the glass tip [3] ; (2) patterns of neural activity with few
or multiple electrode arrays [4–6] ; (3) local fi eld poten-
tials [4, 7] ; (4) electrocorticogram (ECoG) with electrode
grids or stripes sub- or epidurally [8–10] ); all invasive
methods require surgery.
The non-invasive recording of the EEG is the most fre-
quently used method in BCI research. Components most
often used are (a) sensorimotor rhythms (SMR) [11–15] ,
(b) slow cortical potentials [16, 17] , and (c) event-related
BOX 17.1
BCI FOR COMMUNICATION AND PROSTHESIS CONTROL
potentials (ERPs) as a response to sensory, auditory, or
tactile stimulation, namely the P300, a positive defl ec-
tion in the EEG about 300 ms after presentation of rare
target stimuli within a stream of frequent standard stim-
uli [18, 19] , and steady-state visually or somatosensorily
evoked potentials [20, 21] as response to visual or tactile
stimulation between 6–24 Hz [22] .
The acquired signals are digitized and subjected to
a variety of feature extraction procedures, such as spa-
tial fi ltering, amplitude measurement, spectral analysis,
or single-neuron separation [23] . In the following step
a specifi c algorithm translates the extracted features
into commands that represent the users ’ intent. These
commands can either control effectors directly such
as robotic arms or indirectly via cursor movement on
a computer screen to activate switches for interaction
with the environment or to select items, words, or letters
from a menu for communication or to surf the Internet.
CH017.indd 218 5/10/2008 2:06:56 PM