Brain–Computer Interfaces - Index of

Brain–Computer Interface in Neurorehabilitation 163
if simultaneous activation of both hemispheres should be allowed for the activation
of the contralesional hand. Whether cortical reorganization after extended BCI
training will allow the reinstatement of peripheral control remains an open question.
By-passing the lesion by re-growth of new or non-used axonal connections
in the primary motor output path seems highly unlikely. However, a contribution
of fibres from the contralesional hemisphere reaching the contralesional hand to
activate the paralysed hand is theoretically possible. In those cases, a generalisation
of training from a peripheral orthosis or any other rehabilitative device to the
real world condition seems to be possible. In most cases, however, patients will
depend on the activation of electrical stimulation of the peripheral muscles or an
orthosis as provided in the study of Buch et al. Some patients will also profit
from an invasive approach using epicortical electrodes or microelectrodes implanted
in the ipsilesional intact motor cortex, stimulating peripheral nerves or peripheral
muscles with voluntary generated brain activity. A new study on healthy subjects
using magnetoencephalography from our laboratory (see [35]) has shown that with
a non-invasive device such as the MEG with a better spatial resolution than the
EEG directional movements of the hand can be classified from magnetic fields
online from one single sensor over the motor cortex. These studies demonstrate the
potential of non-invasive recordings but do not exclude invasive approaches for a
subgroup of patients non-responding to the non-invasive devices. Also, for real life
use, invasive internalised BCI systems seem to function better because the range
of brain activity usable for peripheral devices is much larger and artefacts from
movements do not affect implanted electrodes. The discussion whether invasive or
non-invasive BCI devices should be used is superfluous: only a few cases select the
invasive approaches, and it is an empirical and not a question of opinion which of
the two methods under which conditions provide the better results.
5 The “Emotional” BCI
All reported BCI systems use cortical signals to drive an external device or a computer.
EEG, MEG and ECoG as well as near infrared spectroscopy do not allow the
use of subcortical brain activity. Many neurological and psychiatric and psychological
disorders, however, are caused by pathophysiological changes in subcortical
nuclei or by disturbed connectivity and connectivity dynamics between cortical
and subcortical and subcortical-cortical areas of the brain. Therefore, particularly
for emotional disorders, caused by subcortical alterations brain–computer interfaces
using limbic or paralimbic areas are highly desirable. The only non-invasive
approach possible for operant conditioning of subcortical brain activity in humans
is functional magnetic resonance imaging (fMRI). The recording of blood-flow and
conditioning of blood-flow with positron emission tomography (PET) does not constitute
a viable alternative because the time delay between the neuronal response, its
neurochemical consequences and the external reward is variable and too long to be
used for successive learning. That is not the case in functional magnetic resonance
imaging, where special gradients and echo-planar imaging allows on-line feedback