Brain-Computer Interface based on Cortical Electrodes

Brainwaves recorded with EEG can be used to control equipment like a computer.
This so-called *Brain-computer Interface* (BCI) technique is currently being
developed to aid patients who can not move or communicate, due to motor-neuron
disease, spinal cord lesion or head trauma. If successful, patients can operate
a computer to write or control devices such as a TV or lights, by merely
thinking about making movements. Currently, the most widely investigated way of
acquiring brainsignals for this purpose is by means of electrodes positioned on
the head, close to the motor cortex. Although it works reasonably, it has
significant drawbacks, including long training schemes, limited performance
(80% good cursor control), and the need for skilled personnel to fit and
operate the EEG system. Recent pilotstudies have shown that much better results
can be obtained with electrodes implanted under the skull, on the cortex. These
studies are sporadic and and lack methodological rigor, as they are performed
ad hoc in patients with electrodes implanted for different (clinical) reasons.
There is a clear need for a proper study, to resolve some key issues relevant
to potential use for paralyzed patients.

The present protocol addresses these issues, with a systematic approach where
we first assess optimal settings for BCI, and then test the quality of BCI
application in adequately sized groups of subjects. The whole project is aimed
at obtaining a sound empirical basis upon which to develop implantable BCI
systems in patients with loss of movement and/or communication.
(original text:) In this project we only conduct the basic research, not the
development of an implantable BCI system.
(amendment 12 jan 2012:) A prototype implantable EEG amplifier will be tested
during planned BCI experiments in the current protocol, for applicability for
paralyzed patients in a future protocol.

We intend to ask them to play computergames while they are monitored at the
IEMU, where they only have to think about specific actions to control the
cursor. We will analyse their brainsignals from electrodes and transform the
signal to pulses to move the cursor. A total of 48 epilepsy patients will be
included, and all testing is conducted in close collaboration with clinical
staff.
Because brain signal from cortical electrodes have excellent qualities,
brainsystems other than the motorsystem can be used for BCI. We will
systematically examine four brain functions: imagined motor movement,
non-vocalized language production, working memory en auditory attention. In
half the patients we will develop algorithms to extract the best signal
features for BCI, in the other half we will test and quantitatively evaluate
the best solution. Before implantation of the grids, patients will be scanned
for brain function (functional MRI), to localize the regions that we know are
involved in the functions of interest. We will then assess whether fMRI is
accurate enough to localize the best brainarea for BCI presurgically.
(amendment 12 jan 2012:) A prototype implantable amplifier will be tested
during already planned BCI experiments in the current protocol. The device is
linked in parallel to the clinical ECoG apparatus and signals from both are
compared offline

(amendement september 2012) A prototype implantable amplifier will be tested
during already planned BCI experiments in the current protocol. The device is
linked in parallel to the clinical ECoG apparatus and signals from both are
compared offline, and the optimal settings for accurate BCI control will be
determined.

The benefit of the study for paralyzed patients is large: it offers them the
possibility to communicate and to control devices using their brain waves. The
study will address several approaches to optimize BCI using ECoG and will give
us important insights for further improvements of BCI systems that are required
for its clinical application. Because Brain Computer Interface applications are
extremely relevant for both adults as well as for children with e.g. spinal
lesions or neuromuscular disorder, we would like to include children > 12 years
old. There are indications that the human brain does not stop developing until
at least early adulthood (Giedd et al., Nat Neurosci 1999; Giedd, Ann NY Acad
Sci, 2004). We would like to investigate whether differences in ECoG
characteristics have any effect on Brain Computer Interface performance. Given
the fact that the UMC Utrecht epilepsy surgery team has had several succesful
7-day grid registrations at the Monitoring Unit in children (as young as 8
years of age), we feel that we can include children of 12 years and older
without imposing an undue burden on them.

The fMRI scan (which is preceded by two short questionnaires, ~5 minutes) will
consist of two sessions of 60 minutes each and will be performed several days
to weeks before the grid implantation. The scans will be planned according to
the patient's wishes, on different days or on one day with a pause in between.
The tasks during the IEMU registration period will last approximately 240
minutes in total, and will be spread over several days, dependent on the
patient's condition, the judgement of the treating physician and the nurse.

During the IEMU registration period, the patients are constantly (24 h/ day)
monitored by a technician and a nurse. In addition, the patients are visited
daily by the treating physician. We will ask the patient for cooperation before
every test or task. The technician, nurse or treating physician can, at any
moment, decide that the test or task cannot be performed by the patient. They
will be instructed to pay specific attention to any sign that indicates that
the patient, at a certain moment, cannot or does not want to, cooperate.

Participation does not form any risk for the adult or minor patients: fMRI
scans are harmless and will be performed according to standard safety
guidelines.

(amendement september 2012): In cases where patients with a vagal nerve
stimulator are sanned on the 3T scanner, a special scan sequence will be used
that has been approved by the MRI safety officer of the UMC Utrecht, and the
vagal nerve stimulator will be tested before and after the san.

Concerning the tasks during the IEMU registration period: Most patients, and
especially children, are extremely motivated to perform the tasks, which makes
the burden nihil. Also the risk is nihil: the computer that is used for BCI is
not directly connected to the patient or to the implanted electrodes. The BCI
computer receives it signal fro mthe ECoG registration computer that records
ECoG continuously for diagnostic purposes. Tests were performed to exclude any
influence of the BCI system on the clinical ECoG system and to make sure that
also a potential crash of the BCI computer does not impose any risk on the
patient.

(amendment 12 jan 2012) In three patients a prototype implantable EEG amplifier
will be tested(outside the body). This device has been tested for safety, and
will be connected in parallel with the clinical EEG system, thus posing no risk
to clinical registration. The test procedure takes place during the already
planned BCI experiments and as such does not increase the burden on the patient.


(amendment september 2012) A prototype implantable EEG amplifier will be
tested(outside the body). This device has been tested for safety, and will be
connected in parallel with the clinical EEG system, thus posing no risk to
clinical registration. The test procedure takes place during the already
planned BCI experiments and as such does not increase the burden on the
patient.