Utrecht Neural Prosthesis (UNP)
A pilot study on controllability of brain signals and application in locked-in patients

In this pilot study we will provide locked-in people with a new means of
communication which has not been possible up to now. For the first time, we
will test whether we can record and decode neural signals obtained directly
from the brain, for control over a computer. The target population is severely
paralyzed, due to trauma, brainstem stroke, neuromotor disease or another
cause, and has no means of communication other than for instance eye blinks.
For these patients there is no technique available to allow them to communicate
unaided. We have developed a system that can read activity directly from the
brain, and can convert the activity to a digital yes/no switch. The system,
called the Utrecht Neural Prosthesis (UNP), consists of an implantable
amplifier for electrical brain signals, a set of electrodes positioned on the
surface of the brain and a wireless receiver (constituting the *Medtronic
System*). This was recently developed by Medtronic, a company specialized in
medical implantable devices such as Deep Brain Stimulators and pacemakers among
others, and has been tested for the current application by our group at the
UMC. A dedicated computer will convert the signals to electrical pulses for
standard Assistive Technology (AT) devices (UMC Brain Interpreter, developed at
the UMC with the Medical Technology Dept). The amplifier and electrodes are
fully implanted and signal is transmitted wirelessly through the skin. Crucial
elements of the UNP are exact positioning of electrodes on the cortex (after
accurate localization of two brain functions and associated regions), and the
specific features extracted from the complex brain signals. The UNP can in
principle enable the patient to engage in any activity that is offered by
commercial Assistive Technology companies that can be performed with yes/no
signals, for instance operating home apparatus such as lights, tv, curtains
etcetera, or writing text for emailing. However, since we do not yet know how
proficient participants can operate the BCI system, the ability to switch home
devices on or off may not be safe. Therefore for the current pilot study we
will limit use of the BCI system to selecting a yes/no indicator on a computer
screen, and to writing text. Most importantly, we aim to achieve unsupervised
function of the BCI, meaning that the patient will be able to use it at home
without the aid of researchers or other experts (but with minimal caregiver
assistance). The device, consisting of amplifier, electrodes and the wires
connecting them to each other, will be implanted in 5 patients over a period of
5 years. Research will be conducted for 12 months after implant and may be
extended each year for 12 months.

The main goal is to achieve a means of communication solely based on brain
signals for severely paralyzed patients. Success will constitute a significant
advance in the field of BCI and of severe paralysis because it will be the
first system that allows for unsupervised operation. For this, we need to
investigate feasibility in the current pilot study. Several objectives are
defined, all representing performance of the BCI system in real life. The
primary objective is to achieve communication via our BCI system in locked in
patients, as measured in terms of writing a sentence on a computer without help
from a member of the BCI research team (unsupervised use). Two secondary
objectives are defined being a) Improve Quality of Life and user satisfaction
with the BCI system, and b) Assess experimental parameters for a larger
clinical study.

This study is an interventional study, lasting 5 years. It is a pilot study
preparing the ground for a larger clinical trial, and has an adaptive trial
design, where we allow for modifications to be made concurrently (e.g.
in/exclusion criteria) or prospectively (e.g. discontinuation following interim
evaluation, modification of end-of-study date), in communication with DSMB and
METC committee (Thabane et al., 2010).

A device will be implanted, to detect and analyze brain signals. After the
surgical implant procedure, feedback is given of brain activity via a visual
display. Successful control over the brain signal will improve a patient*s
wellbeing since it offers a means of communicating. Failure to control the
signal will induce disappointment. No detrimental effects on physical or mental
health are to be expected.

General
The research is fully directed at the patient population participating in the
study. The participating patients are likely to experience benefit from
participation in terms of acquiring a new means of communication for the
duration of the study. If the study succeeds in its objectives, it can provide
patients with the means to engage in interaction with others and with their
environment without the help of others, and at any time. This degree of
autonomy is not available in any other way for these patients.
However, the research is high risk, in spite of the significant amount of
proof-of-principle research leading up to this study, because there are no data
from chronic implants yet that would allow us to estimate performance of the
BCI system in real life. The current medical device is the first to be
implanted for the purpose of BCI worldwide so there are no data to further
strengthen the proof of principle research. There is a risk that the system
fails due to unforeseen phenomena. Nevertheless, the prior research results
strongly suggest that the experiment should work well in at least 50 % of the
participants, and reasonably well in the other 50%, provided that electrodes
are in the proper position (Vansteensel et al., 2010).
The participants will spend time in the UMC for the procedure, which involves
two surgeries: one to position sets of electrodes and their leads, and one
three days later to place the device and connect the leads. During their stay
they undergo testing for determining the optimal electrode pairs for use, and
start practising BCI with the signals. In case a participant already has ECoG
electrodes and Activa PC+S implanted, one of the surgeries and the
inter-surgery testing will not take place and its associated risks and burden
do not apply. After recovery the patients go home and training continues until
they can operate an AT device (Touchy). They continue to participate in
research sessions aimed at obtaining performance measures, measures of use and
satisfaction, and at improving the decoding technique to further improve
performance (less error, faster switching). Hence the burden will be medical
(surgical procedure), and devoting time and energy to the experiment. If the
correct decoding of brain signals fails, and a patient is not able to gain
accurate control over the BCI, this patient has no benefit of contributing to
the study. If, however, correct decoding is successful, the benefit is possibly
large, since he or she will have a new means of communicating with others and
control of home environment.
After implantation, MRIs and the use of diathermia are no longer possible. If,
during the study, an MRI is required for urgent medical reasons, the implanted
parts need to be surgically removed before the scan.
Risks include those related to the fMRI scan, surgery, technical aspects of the
device and emotional wellbeing. These risks are discussed below.

fMRI*
fMRI scanning is considered a safe procedure, and potential risks will be
minimized by excluding patients who do not meet the inclusion/exclusion
standards. Special care is taken to ensure that the patient has no metal
objects present in the body. Of particular concern are metal fragments in brain
tissue or eyes, surgical clips and non-removable electronic devices, such as
pacemakers. Potential risks associated with the transport of a patient on
artificial ventilation from the ICU to the MRI suite, and switching to the MRI
compatible respiratory aid equipment in the MRI suite, are minimized by
following the Standard Operation Procedure (see Section 8.3.3 and 15.1). The
patient*s vital signs such as heart rate, respiratory rate and oxygen
saturation will be monitored continuously by the anesthesiology team.
Furthermore, a member of the research team will be monitoring the available
communication channel with the patient for the total duration of the fMRI scan.
In the case of unreliable communication, this monitoring will not be possible
possible and the vital signs of the patient will be informative about the
patient*s wellbeing. In consultation with the caretaker, one or more
interruptions of the scan session may be scheduled for extra checkups and care
if needed.
* Because electrodes are already in place, imaging, electrode placement surgery
and strip selection procedures are not relevant for new participants with
existing implant


Surgery
The experiments involve a surgical procedure with implantation of electrodes
underneath the skull, lead wires under the skin from the head to the chest,
and/or positioning of a medical device under the skin on the chest. This
procedure is the same as for deep-brain stimulators, but with a less invasive
electrode positioning as they do not penetrate brain tissue (deep-brain
stimulators travel through much of the brain white matter to reach their goal
in the thalamus). Medtronic has already delivered 85.000 DBS stimulators for
implantation in Parkinson patients alone, indicating that implantation of the
device is by now a standard procedure.
The surgical implantation procedures (implantation of the electrodes and
implantation of the Activa PC+S device) carry the same risks associated with
any other brain surgery. In the worst case, risks of brain surgery may include
serious complications such as coma, bleeding inside the brain, seizures and
infection. Some of these may be fatal. However, the chance that complications
occur is very low, among others since electrodes do not penetrate the cortex.
The risk of subdural or epidural hemorrhage ranges from none to 2%. This
literature is mostly based on devices penetrating the cortex (Bronstein et al.,
2011; Franzini et al., 2011). Once implanted, the system may become infected.
The risk of intracranial infection after implantation surgery is around 5% in
the literature. The risk of infection at the scalp where the cables are
externalized is likely smaller. Depending on the severity and the location of
the infection, parts of the system may have to be explanted. After
implantation, parts may wear through the skin, and the lead or lead/extension
connector may move. Any of these situations may require additional surgery.
Notably, having two brain surgeries close together in time is standard clinical
practice in the treatment of severe intractable epilepsy and other conditions,
and does in itself not add any additional risks.
Risks related to the explantation scenarios (see Section 8.3.12) include
infection (~3%) and, in the case of explantation of all implanted parts
(including the electrodes), subdural or epidural hemorrhage (0-2%).

Technical aspects of the device
In case the UNP does not work (properly), or stops working after a period of
good performance, the source of the problem has to be identified. There are a
number of technical aspects or causes that need to be considered:
- The brain signal may show unexpected characteristics, making it an unreliable
source of signals for UNP control. Based on our prior research with epilepsy
patients we expect that the chance that this happens is very small, especially
in the first period after surgery. The longterm effects of the use of a certain
brain area for controlling the UNP on the characteristics of the measured
signal is, however, presently unknown and one of the parameters investigated in
the current study. Evidence from a recent study looking at action potentials
indicates that five years after implant signals are still usable for BCI
purposes (Hochberg et al. 2012). If the characteristics of the primary control
signal (see Section 8.3.7) prove to be too unreliable for correct UNP
performance, we will switch to the secondary control signal and attempt to
achieve UNP control using that.
- One of the external parts (outside the body) of the UNP is defective. In this
case, the broken part may be repaired or replaced.
- One of th