You are on page 1of 8

Literature Review

BraineMachine Interfaces: The Role of the Neurosurgeon


Aswin Chari1,2, Sanjay Budhdeo3,4,6, Rachel Sparks7, Damiano G. Barone8, Hani J. Marcus5,9, Erlick A.C. Pereira10,
Martin M. Tisdall1,2

Key words Neurotechnology is set to expand rapidly in the coming years as technological
- Bioethics innovations in hardware and software are translated to the clinical setting.
- Brainecomputer interface
- Brainemachine interface
Given our unique access to patients with neurologic disorders, expertise with
- Microelectrode which to guide appropriate treatments, and technical skills to implant braine
- Neuroprostheses machine interfaces (BMIs), neurosurgeons have a key role to play in the
- Neurotechnology
progress of this field.
Abbreviations and Acronyms We outline the current state and key challenges in this rapidly advancing field,
BMI: Brainemachine interface including implant technology, implant recipients, implantation methodology,
DBS: Deep brain stimulation implant function, and ethical, regulatory, and economic considerations. Our key
FDA: Food and Drug Administration
LFP: Local field potential
message is to encourage the neurosurgical community to proactively engage in
collaborating with other health care professionals, engineers, scientists,
From the 1Developmental Neurosciences, Great Ormond ethicists, and regulators in tackling these issues. By doing so, we will equip
Street Institute of Child Health, University College London,
London, United Kingdom; 2Department of Neurosurgery,
ourselves with the skills and expertise to drive the field forward and avoid being
Great Ormond Street Hospital, London, United Kingdom; mere technicians in an industry driven by those around us.
3
Department for Clinical and Movement Neurosciences,
Queen Square Institute of Neurology, University College
London, London, United Kingdom; Departments of
4
Neurology and 5Neurosurgery, National Hospital for novel BMI package comprises 1024 not just restorative and functional but also
Neurology and Neurosurgery, London, United Kingdom; contacts in custom-built microelectrode augmentative neurosurgery.
6
OwkinInc, New York, New York, USA; 7School of Biomedical threads, implanted into the brain by a In this article, we outline the current
Engineering and Imaging Sciences, King’s College London,
London, United Kingdom; 8Division of Neurosurgery,
robotic system, and is able to wirelessly state and key challenges in this rapidly
Department of Clinical Neurosciences, University of transmit these signals in real time.1 advancing field, including implant tech-
Cambridge, Cambridge, United Kingdom; 9Wellcome EPSRC Although this news has received a lot of nology, implant recipients, implantation
Centre for Interventional and Surgical Sciences, University public interest, many components of the methodology, implant function, implant
College London, London, United Kingdom; and
10 proposed technology are not regulation, and ethical, regulatory, and
Neurosciences Research Centre, Molecular and Clinical
Sciences Research Institute, St George’s, University of groundbreaking; systems with similar economic considerations. Our key mes-
London, United Kingdom capabilities have been reported in the peer- sage is to encourage the neurosurgical
To whom correspondence should be addressed: reviewed literature as long as 17 years ago fraternity to proactively engage in collab-
Aswin Chari, M.A., B.M.B.Ch. and other simpler systems are used clinically orating with other health care pro-
[E-mail: aswinchari@gmail.com] to treat a variety of neurologic disorders.3 fessionals, engineers, scientists, ethicists,
Citation: World Neurosurg. (2021) 146:140-147. Neuromodulation technology, including and regulators in tackling these issues. By
https://doi.org/10.1016/j.wneu.2020.11.028 deep brain stimulation (DBS), is already a doing so, we will equip ourselves with the
Journal homepage: www.journals.elsevier.com/world- mature market worth more than U.S. $5.8 skills and expertise to drive the field for-
neurosurgery
billion in 2020 and set to expand rapidly in ward responsibly and avoid being mere
Available online: www.sciencedirect.com the coming years as technological in- technicians in a field driven by those
1878-8750/$ - see front matter ª 2020 Elsevier Inc. All novations are translated to the clinical around us.
rights reserved.
setting, with 1 report forecasting a world-
wide market of $13.3 billion by 2022.4-6
INTRODUCTION Neurosurgeons have a key role to play in IMPLANT TECHNOLOGY
Elon Musk’s August 2020 press conference its progression because we have a unique Clinical indications should be considered in
outlining the progress of his new braine relationship with patients affected by the context of neural interfaces that are in
machine interface (BMI) company, neurologic disorders that may benefit from use and those that are in development.
Neuralink, captured the attention of neuro- BMIs, both as treating physicians and in Broadly, existing devices that interface with
scientists and technology enthusiasts guiding their decision-making processes as the brain can be divided into those that
worldwide because he showed the ability to to the best choice of treatment. We also record or stimulate neural activity (Figure 1).
record neurons from pig cortex in real time. possess the skills and expertise to implant Recording devices include macroelectrodes
He had earlier promised to “merge” humans these new devices into the nervous system. such as stereoelectroencephalography
with artificial intelligence when he first It is therefore easy to see how many neu- electrodes or cortical grids/strips and
announced the company.1,2 Neuralink’s rosurgeons may be part of a subspecialty of microelectrode arrays; there are

140 www.SCIENCEDIRECT.com WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2020.11.028


LITERATURE REVIEW
ASWIN CHARI ET AL. BRAINeMACHINE INTERFACES: THE ROLE OF THE NEUROSURGEON

adaptations of recording macroelectrodes Although not shown in the recent pre- the nervous system such as spinal cord,
with microelectrode contacts, but these print, novel BMI systems such as those peripheral nerves, and cranial nerves
devices are used only in research proposed by Neuralink are designed to be (including cochlear implants and vagal
settings.7,8 An endovascularly implantable precision systems with hundreds to thou- nerve stimulation).16,25 Although an
stent electrode that is placed in the cortical sands of electrodes that allow in-depth exploration of these specific
vasculature has recently received programmed stimulation at each contact.1 technologies is beyond the scope of this
breakthrough device designation from the Although hypothetical, this technology review, there is a lot of cross-fertilization
U.S. Food and Drug Administration (FDA) may, in time, allow the individual to of technological breakthroughs and
and is undergoing clinical trials. sense (somatic sensation, vision, smell, mechanistic insights across these different
Stimulating devices are mostly in the form or taste) using external sensors or allow modalities.
of DBS electrodes, although other the device to alter brain connectivity, There are several key areas of research
constructs, such as auditory brainstem affecting cognitive, psychological, and regarding improving this technology. The
implants, also exist.9,10 Novel constructs emotional responses. first is the foreign body reaction, a classic
such as closed-loop DBS and responsive Devices that record neural activity can physiologic response of the body to
neurostimulation electrodes are capable of be divided into macroelectrodes that implanted foreign material.26 In the context
both recording and stimulating, with the measure local field potential (LFP) activity of BMIs, this reaction affects both the
aim of optimizing stimulation in real time aggregated from many neurons and mi- short-term and long-term performance of
based on the activity recorded.11,12 The croelectrodes that are capable of the recording and stimulation capabilities
choice of device largely depends on the measuring extracellular action potentials of the device because the formation of
indication and the location of recording/ from single neurons in addition to LFP. fibrotic tissue around the interface causes
stimulation; some devices are better suited Both are immensely powerful when com- an inefficient transduction of the electric
to record and stimulate cortical structures, bined with modern data-processing and signal.27,28 Many factors have been
whereas others are more suited to deeper machine-learning technologies; cortical associated with the degree of foreign body
brain structures. LFP signals can decode speech and exist- reaction, including surface properties of
In terms of stimulating or modulating ing microelectrode arrays have been used the biomaterial (porosity, roughness,
brain activity, DBS strategies offer an to control a range of functions, including stiffness, and chemistry), shape, surface
ability to modulate the pulse current, prosthetic arm and cursor control, in area and volume of implant, degree of
duration, and frequency to a small number small numbers of patients.18-22 Like with implantation trauma, and mechanical
of electrodes (usually 4e16) in a specific the stimulating electrodes, these devices mismatch between the implanted stiff
area of the brain. Targets are chosen based are limited by the number of electrodes material and the soft biological tissue.27-30
on the specific indication and patient and sampling from a small area of brain; Novel biocompatible implants have shown
characteristics. Treatment is optimized by current microelectrode arrays have in the our ability to record microelectrode
manually titrating the settings to the region of 100 electrode contacts and activity from large numbers of channels
clinical response. Novel strategies that are sample a 1-cm2 area of cortex. Newer de- for up to 6 months in animals.31,32
used to improve DBS include directional vices may be able to sample from thou- Clinically, this area must be approached
electrodes, closed-loop systems that are sands or tens of thousands of neurons but with caution, warning patients that
able to record and stimulate, and con- the advantages of recoding from positive effects may diminish over time
nectomic strategies in which individual increasing numbers of neurons have yet to and render implants ineffective. In
patient structural connectivity is used to be realized.23 Implanting hundreds of addition to the basic science work that is
optimize target location at the time of microscale biocompatible wires into being undertaken to understand the
implantation.12,13 DBS and responsive eloquent tissue also requires careful mechanisms of the foreign body reaction
neurostimulation technology have a consideration of risks. Despite the small and options for subverting it, we suggest
profound impact on people with scale, implanting microelectrodes into establishing rigorous implant registries to
movement disorders, epilepsy and, the eloquent cortex has been shown to determine longer-term durability in
increasingly, psychiatric cause fine motor deficits in animal humans. Other issues that warrant study
disorders.9,11,12,14,15 In addition to DBS, models and the long-term impact of this include the impact of electrode drift on the
emerging stimulation techniques for requires evaluation.24 Electrodes may fidelity of the captured signals and the
auditory and visual restoration hold preclude or cause artifact on subsequent long-term impact of neural implants on
promise; although auditory brainstem imaging, potentially interfering with brain connectivity and function.
implants have been in clinical use for diagnostic accuracy and subsequent
many years, several research groups are medical treatment. Magnetic resonance
investigating broadly similar strategies imagingecompatible neuromodulation IMPLANT RECIPIENTS
for visual cortex stimulation based on devices are entering the market, but Given our ability to record and stimulate
information from a camera mounted on further work is required for specific BMI neural activity, it is not surprising that in-
the forehead or glasses.10,16,17 implants. dications for BMIs include a wide range of
Stimulating technology is limited by In addition to implants that interface neurologic and psychiatric disorders that is
small numbers of electrodes that are with the brain, neural interface technology constantly expanding. There are approved
spatially limited to small structures. may also interface with other elements of indications for DBS in Parkinson disease,

WORLD NEUROSURGERY 146: 140-147, FEBRUARY 2021 www.journals.elsevier.com/world-neurosurgery 141


LITERATURE REVIEW
ASWIN CHARI ET AL. BRAINeMACHINE INTERFACES: THE ROLE OF THE NEUROSURGEON

Figure 1. Scope of brainemachine interfaces that splits the field into recording and stimulating devices. DBS, deep brain stimulation; RNS, responsive
neurostimulation.

essential tremor, dystonia, and obsessive- introduced into the clinical domain; we an individual and societal level. As medi-
compulsive disorder and emerging suggest that these are undertaken solely cal professionals and key members of the
indications in epilepsy, neurocognitive under the auspices of a clinical trial using BMI community, neurosurgeons need to
disorders, pain, and other neuropsychiatric structured frameworks for the introduc- think carefully about the medical, ethical,
disorders.9,33 More experimental tion of new technology with adequate and societal implications of this situation
indications for BMIs include controlling regulation and oversight.35,36 and, importantly, whether and how we
prostheses,19-21 obesity, multiple sclerosis, In addition to medical indications, should be involved in such practices,
substance addiction, and memory BMIs hold immense potential to augment especially in the context of health care
augmentation/editing.34 function (e.g., memory, cognition, sensa- systems with limited resources.
Determining which patients are eligible tion, language, motor control) in other-
to receive implants is an individualized wise normal individuals.37 Although this is
riskebenefit analysis, often undertaken by not be the focus of current research and IMPLANTATION METHODOLOGY
a multidisciplinary team consisting of may be seen as unethical by some, Perhaps the most straightforward challenge
neurologists, neurosurgeons, neuroradi- augmenting function is a natural in this field involves achieving accurate,
ologists, psychiatrists, and allied health corollary of developing technology for safe, and long-lasting implantation of
professionals who weigh the risks of sur- functional restoration in those with electrodes. This challenge has been a key
gery and implant maintenance against the neurologic disorders. For example, driver behind neurosurgical technology for
probability of clinical improvement. Fac- noninvasive sensors and stimulators have decades, starting with frame-based stereo-
tors that are taken into consideration already been used to achieve direct brain- tactic localization based on air ven-
include disease severity, associated to-brain communication38 and it is triculography and resulting in the modern
comorbidities, imaging abnormalities, possible that invasive strategies will only plethora of robot-assisted neuronavigation
and, significantly, a lot of importance is increase precision of such systems. If systems that incorporate high levels of
given to patient preference. Especially and when such a situation arises, careful submillimeter accuracy and integrate with
when considering novel or experimental consideration must be given to the riske advanced vascular imaging to ensure that
indications, careful consideration must be benefit balance in the absence of disease blood vessels are avoided.39 These systems
given to the way in which these are and what level of risk is acceptable, both at have all been supervisory control systems,

142 www.SCIENCEDIRECT.com WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2020.11.028


LITERATURE REVIEW
ASWIN CHARI ET AL. BRAINeMACHINE INTERFACES: THE ROLE OF THE NEUROSURGEON

which require human input to control; on training requirements, and national incorporation of novel approaches such
there is scope for fully automated systems and international implant registries will as network science that provides a
that implant autonomously, which raises aid ongoing audit and oversight of efficacy natural language to model the complex,
further issues such as responsibility and and complication rates. changing system of the brain.51 Although
liability, which are actively being there is an increasing wave of clinician-
explored.40 Microelectrode arrays and scientists who have the computational
depth electrodes require cortical IMPLANT FUNCTION knowledge to be able to design and deploy
penetration. Histologic analyses from As mentioned earlier, recording electrodes these algorithms themselves, dealing with
microelectrode arrays, implanted largely can be divided into traditional macro- such data and the code behind their pro-
in research contexts during short-term electrodes and novel microelectrodes cessing is probably best served by close
monitoring of patients with epilepsy, depending on the size of the recording collaboration with computational neuro-
confirm minimal tissue damage associated contacts. Microelectrodes record at fre- scientists, engineers, and mathematicians.
with pneumatic implantation devices quencies around 30 kHz and capture mul-
designed to minimize trauma, but im- tiscale electrophysiologic data, including
plantation is not without risk.41,42 Surgical LFP (1e100 Hz), high-frequency oscilla- IMPLANT REGULATION AND
techniques must therefore be constantly tions (80e500 Hz), and extracellular action MONITORING
evaluated and rigorously audited to ensure potentials (>300 Hz) from neurons; these
that the highest standards are maintained. frequency bands are arbitrary and have Implant Monitoring
A more complicated challenge in im- been determined by identified features of Two specific challenges in this area include
plantation is accurately identifying the interest in each. In epilepsy monitoring, dealing with large volumes of data, data
appropriate region of the brain to target. microelectrodes have been shown to cap- security, and ownership. Microelectrode
Traditionally, implantations have focused ture neural signals that are not captured by recordings generate significant volumes of
on anatomic structures that can be visual- clinical macroelectrodes.8,48 Extracellular data (250 channels at 30 kHz is approxi-
ized on magnetic resonance imaging. For action potentials are recorded from a few mately 115 GB per hour). This amount of
instance, DBS for the treatment of Parkin- microelectrode contacts and require spike data is difficult to store and process on
son disease is targeted toward the sub- sorting, a computationally intensive many of the computer systems in place in
thalamic nucleus or globus pallidus process that ascribes particular action many hospital systems. However, develop-
internus. However, the true functional potential waveforms to putative neurons.49 ment of high-performance computing
target is the motor subset of these struc- Although significant progress has been systems and cloud-based computing may
tures, and additional information from made in spike sorting algorithms, the provide a solution that can scale with ever-
microelectrode recordings or advanced fidelity of online (real time) spike sorting increasing demands on data storage and
imaging techniques is necessary for accu- over longer time scales than a few hours computation.52,53
rate placement.43-45 However, as more has not been established. Even if spikes Data security is of utmost concern
complex disorders are treated, especially cannot be sorted, much can be gleaned because altered functioning or disabling of
those related to mood or cognition, iden- from unsorted multiunit activity.3 Factors neuroimplants can have devastating con-
tifying the appropriate target becomes such as electrode drift and signal decay sequences. System vulnerabilities may be
more complex and may require a combi- from the foreign body reaction need to be exploited, leading to malicious alterations
nation of advanced structural and evaluated. Action potentials from single to inbuilt algorithms, termed brainjack-
functional imaging techniques and elec- neurons can then be analyzed in several ing.54 Furthermore, even without malicious
trophysiology to help guide preoperative ways, including the rate of firing, timing intent, we must safeguard against
assessment. of firing in relation to the underlying LFP inadvertent access, which can be caused
Novel constructs, especially those phase, and population firing of multiple by the user, interactions with other
seeking to record macroscale signals, may neurons. systems (wireless networks and hardware
be implanted through endovascular routes, Historical understanding of these elec- that create electromagnetic signals), or
with the ability to record and stimulate trophysiologic signals has been limited to even errors during desired software
when implanted into cortical vessels. This pattern recognition on visual inspection upgrades. Most devices avoid data security
construct would preclude the need for a but progress in computational power has issues by acting as a closed system, in
craniotomy although it would be limited by enabled the application of signal process- which information is not stored externally
the location of cortical vessels. One such ing tools to better understand them, to the and can be adjusted only in person.
device is undergoing first-in-human trials extent that we are developing atlases of Although there are wireless implantable
and has received breakthrough device normal intracranial electroencephalo- devices on the market, they are typically
designation from the FDA.46,47 graphic dynamics.50 Machine-learning al- secured using external relay devices, in
It is possible that some future neuro- gorithms that are being applied to these which a physical object is required to be
surgeons will be implant neurosurgeons signals will need to incorporate data from placed near the device to gain access to
and we also need to adapt our curricula to all these scales to optimize the output of the signal.55 However, such a system
equip future surgeons with the required BMIs, a field that is in its relative infancy necessarily limits the ability to monitor
technical and nontechnical skills. but that has shown immense promise.18 signals in real time and make
Specialist societies must issue guidance These algorithms may benefit from simultaneous adjustments. To unlock this

WORLD NEUROSURGERY 146: 140-147, FEBRUARY 2021 www.journals.elsevier.com/world-neurosurgery 143


LITERATURE REVIEW
ASWIN CHARI ET AL. BRAINeMACHINE INTERFACES: THE ROLE OF THE NEUROSURGEON

can be interpreted through BMIs as


Table 1. New Human Rights Principles in the Era of Neurotechnology and showing risk of public danger, society must
Neuroprostheses51 accept trade-offs between autonomy, pri-
Cognitive liberty The right to alter one's mental state with the help of neurotechnology as well as to vacy, and public security. However, the use
refuse to do so of such devices to tailor marketing cam-
paigns or other commercial activities
Mental privacy The right to one's own brain data. It should not be recorded, shared, or used without
explicit consent should be safeguarded against in the in-
terest of the patient. Situations may arise
Mental integrity Organizations and governments should not alter the computation of the brain without involving employers and insurance com-
consent
panies mandating such implants, as has
Psychological Personal identity should not be compromised already been seen in peripherally implanted
continuity microchips.61
Stimulating devices raise issues of au-
tonomy; although they have the ability to
potential, systems must be designed to intelligenceebased algorithms are rapidly increase the functional independence of
ensure that access is limited to only being developed to improve recording and those with progressive neurologic disorders,
trusted vendors. closed-loop technology, robust evaluation is the individual still self-governing and to
Data protection is also a key consider- of these novel algorithms is crucial, as are what extent are they still accountable for
ation, both in terms of outright theft and other aspects such as the data used to their actions?60,62 By extension, ethicists
ensuring that data are used only for their assess them. Existing frameworks such as have debated whether BMIs become part of
intended purposes. Frameworks must the FDA Software as a Medical Device are the body schema and integrate into the
clearly delineate who owns the data, who is being adapted for artificial intelligence person, both from legal and philosophic
responsible for their safe storage, where and machine-learning algorithms56 and standpoints.62 Augmenting function in
they are stored, and the rights of the indi- may require further modification specific otherwise healthy individuals also raises
vidual, medical professionals, companies for neurotechnology and BMIs. Although issues of risk and societal implications,
and governments to access, use, and patient data must be sufficiently with entrenched and widening social
monetize such data. Existing regulations, protected, open science and open data inequalities between those who can and
such as General Data Protection Regula- sets have hastened progress in recent cannot afford such implants.
tion, may largely cover these requirements. years. Regulations must balance both Bioethicists have identified that current
sides and novel constructs such as data human rights principles may be insufficient
obfuscation should be used to allow for dealing with the advances in neuro-
Implant Regulation pretraining of machine-learning algo- technology and have identified 4 new guid-
New active implantable medical devices rithms and preserve confidentiality.57 In ing principles (Table 1).63 What is
and their accompanying software require circumstances in which decision making particularly interesting about this
careful evaluation both before and during is difficult, citizens’ juries may be used framework is the aspect of cognitive liberty
human trials. Although an efficient to decide whether data may be shared that gives people a right to alter their
approval process is crucial for clinical with other parties for research and mental state. If neurosurgeons are to be
translation and patient benefit, this must commercial purposes.58 involved in altering the mental state of
not occur at the cost of robust evaluation otherwise normal individuals, we must
of the clinical efficacy and risks. Estab- think carefully about the levels of
lished approval processes for medical de- ETHICAL CONSIDERATIONS acceptable risk, informed consent
vices (CE [Conformité Européenne] marking The merging of humans with machine frameworks that protect both individuals
and FDA approval) need to be adapted and interfaces (potentially with a super- and ourselves, and, in the context of
expanded to increase the quality of imposed machine-learning application research evaluations, our posttrial
ongoing robust evaluation of new tech- layer) raises serious ethical issues.59,60 responsibilities to the participants and the
nologies, specifically to consider carefully First, issues of consent that apply to public.
individual and population level thresholds individuals without capacity and children
for riskebenefit considerations, in which would continue to be relevant, with the
indications for invasive stimulation are for added burden of the psychological ECONOMIC CONSIDERATIONS
wellness or augmentation of physiologic impact of BMIs.60 Economic considerations can be broken
function. Frameworks such as IDEAL-D, Devices that record from the brain may down into 3 pertinent questions. The first
which seek to end the dichotomy of (intentionally or inadvertently) have access question is who will develop BMIs. The
approved versus not approved, must be to private or intimate thoughts not meant possible archetypes of organization are
adopted and we, as the responsible clini- for the public world, an issue both in terms universities, hospitals, nonuniversity state-
cians, must champion these approaches.35 of recording and storage of such informa- owned research facilities, and small and
Software development is likely to play a tion. There may be questions of the right large companies. At different times,
pivotal role in the neurotechnology and extent to which privacy should be different organizational archetypes may be
sphere. Although network and artificial preserved in these situations; if thoughts best placed to deliver on different steps

144 www.SCIENCEDIRECT.com WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2020.11.028


LITERATURE REVIEW
ASWIN CHARI ET AL. BRAINeMACHINE INTERFACES: THE ROLE OF THE NEUROSURGEON

Figure 2. Framework outlining some of the key challenges for the brainemachine interface community, highlighting areas in which active clinician involvement
is crucial to progress.

such as ideation, productization, adher- funding may come with the expectation of range of medical and nonmedical uses.
ence to regulatory standards, quality con- return on invested capital.65 We have not even considered the vast
trol and assurance, business development, The third question is who will pay for range of noninvasive stimulation strategies
and marketing. An important consider- BMIs. For BMIs that are developed to that will sit alongside invasive implants,
ation is the incentives for each type of enable return to function for patients who some of which have been commercialized
organization, including scientific prog- have lost abilities that they previously had, for improving cognitive performance.66
ress, betterment of health, and profits. or might reasonably be expected to have, Clinicians, and specifically neurosur-
Irrespective, progress must be evaluated the starting point will be existing payor geons, hold a unique position in this field
objectively, and all organizations must be mechanisms, such as governments or in- because our skill set makes us gatekeepers
held to the same exacting standards. As surance companies, depending on health to the clinical application of such technol-
neurosurgeons, we hold a unique position economy. For augmentation of normal ogy. We must therefore take leadership
to generate insights into product devel- function, our assumption is that the pay- roles in shaping the field. We need to
opment and usefulness; clinicians can play ment burden will be on the individual. This continue working closely with the engi-
important roles in developing relevant or- situation has the potential to exacerbate neering and computational neuroscience
ganizations.64 One concern we must and entrench existing inequalities within communities to improve implant materials,
therefore keep in mind is potential society and place neurosurgeons in a po- minimize the foreign body reaction, ensure
conflicts of interest that might arise sition of ethical dilemma when considering that surgical implant techniques minimize
because of financial interests in participation in such activities. risk and optimize efficacy, and optimize
commercial organizations. algorithms for understanding and stimu-
The second question is who will fund the lating the brain. Given the rapid pace of
development. Funding may be institutional CONCLUSIONS technological advancement, we also need
(governmental, nongovernmental, or pri- The road ahead for the BMI community is to be involved in preemptively shaping the
vate) or deployed capital from venture long, both in terms of technological legislative and policy agenda to ensure that
capital firms. The scales of money available innovation and ethical and moral consid- such technology is introduced with
through these different routes are likely to erations. There is no certainty that Elon adequate regulation and is used for ethical
be vastly different. For example, Neuralink Musk and Neuralink will provide the indications. Some of the key challenges for
has amassed more than U.S. $150 million breakthrough in this field but there is a the BMI community, highlighting areas in
of funding in its short history but such clear direction of travel with an increasing which active clinician involvement is crucial

WORLD NEUROSURGERY 146: 140-147, FEBRUARY 2021 www.journals.elsevier.com/world-neurosurgery 145


LITERATURE REVIEW
ASWIN CHARI ET AL. BRAINeMACHINE INTERFACES: THE ROLE OF THE NEUROSURGEON

to progress, are outlined in Figure 2. 13. Rodrigues NB, Mithani K, Meng Y, Lipsman N, nerve implanted electrode. Conf Proc IEEE Eng Med
Hamani C. The emerging role of tractography in Biol Soc. 2010;2010:1543-1546.
National and international neurologic and
deep brain stimulation: basic principles and cur-
neurosurgical bodies should lead the rent applications. Brain Sci. 2018;8:23. 30. Veiseh O, Doloff J, Ma M, et al. Size- and shape-
charge in setting up task forces for these dependent foreign body immune response to
purposes. We, the neurosurgical 14. Shanechi MM. Brain-machine interfaces from materials implanted in rodents and non-human
motor to mood. Nat Neurosci. 2019;22:1554-1564. primates. Nat Mater. 2015;14:643-651.
community, must engage to avoid
becoming mere technicians in this rapidly 15. Bari AA, Thum J, Babayan D, Lozano AM. Current 31. Chung JE, Joo HR, Smyth CN, et al. Chronic im-
advancing field. and expected advances in deep brain stimulation plantation of multiple flexible polymer electrode
for movement disorders. Prog Neurol Surg. 2018;33: arrays. J Vis Exp. 2019;152:10.3791/59957.
222-229.
REFERENCES 32. Chung JE, Joo HR, Fan JL, et al. High-density,
16. Niketeghad S, Pouratian N. Brain machine in- long-lasting, and multi-region electrophysiolog-
1. Musk E, Neuralink. An integrated brain-machine terfaces for vision restoration: the current state of ical recordings using polymer electrode arrays.
interface platform with thousands of channels. cortical visual prosthetics. Neurotherapeutics. 2019; Neuron. 2019;101:21-31.e5.
bioRxiv. Published online August 2, 2019:703801 16:134-143.
https://doi.org/10.1101/703801. 33. Hariz M, Blomstedt P, Zrinzo L. Future of brain
17. Wong K, Kozin ED, Kanumuri VV, et al. Auditory stimulation: new targets, new indications, new
2. Crane L. Elon Musk demonstrated a Neuralink
brainstem implants: recent progress and future technology. Mov Disord. 2013;28:1784-1792.
brain implant in a live pig. New Scientist. Available
perspectives. Front Neurosci. 2019;13:10.
at: https://www.newscientist.com/article/2253274-
34. Youngerman BE, Chan AK, Mikell CB,
elon-musk-demonstrated-a-neuralink-brain-implant- McKhann GM, Sheth SA. A decade of emerging
18. Anumanchipalli GK, Chartier J, Chang EF. Speech
in-a-live-pig/. Accessed August 31, 2020.
synthesis from neural decoding of spoken sen- indications: deep brain stimulation in the United
tences. Nature. 2019;568:493-498. States. J Neurosurg. 2016;125:461-471.
3. Carmena JM, Lebedev MA, Crist RE, et al.
Learning to control a brainemachine interface for
19. Jarosiewicz B, Sarma AA, Bacher D, et al. Virtual 35. Sedrakyan A, Campbell B, Merino JG, Kuntz R,
reaching and grasping by primates. PLoS Biol.
typing by people with tetraplegia using a self- Hirst A, McCulloch P. IDEAL-D: a rational
2003;1:e42.
calibrating intracortical brain-computer interface. framework for evaluating and regulating the use of
Sci Transl Med. 2015;7:313ra179. medical devices. BMJ. 2016;353:i2372.
4. Research and Markets. The Market for Neuro-
technology: 2018-2022. Available at: https://www.
20. Wodlinger B, Downey JE, Tyler-Kabara EC, 36. McCulloch P, Altman DG, Campbell WB, et al. No
researchandmarkets.com/reports/4636680/the-market-
Schwartz AB, Boninger ML, Collinger JL. Ten- surgical innovation without evaluation: the IDEAL
for-neurotechnology-2018-2022. Accessed January
dimensional anthropomorphic arm control in a recommendations. Lancet. 2009;374:1105-1112.
15, 2020.
human brain-machine interface: difficulties, so-
5. Marcus HJ, Hughes-Hallett A, Kwasnicki RM, lutions, and limitations. J Neural Eng. 2015;12: 37. Mankin EA, Fried I. Modulation of human mem-
016011. ory by deep brain stimulation of the entorhinal-
Darzi A, Yang G-Z, Nandi D. Technological
innovation in neurosurgery: a quantitative study. hippocampal circuitry. Neuron. 2020;106:218-235.
J Neurosurg. 2015;123:174-181. 21. Gilja V, Pandarinath C, Blabe CH, et al. Clinical
translation of a high-performance neural pros- 38. Jiang L, Stocco A, Losey DM, Abernethy JA,
6. Neuromodulation Market | By Technology & Appli- thesis. Nat Med. 2015;21:1142-1145. Prat CS, Rao RPN. BrainNet: A multi-person
cation |MarketsandMarkets. Available at: https:// brain-to-brain interface for direct collaboration
www.marketsandmarkets.com/Market-Reports/neuro 22. Makin JG, Moses DA, Chang EF. Machine trans- between brains. Sci Rep. 2019;9:1-11.
stimulation-devices-market-921.html. Accessed lation of cortical activity to text with an encodere
August 24, 2020. decoder framework. Nat Neurosci. 2020;23:575-582. 39. Guo Z, Leong MC-W, Su H, Kwok K-W, Chan DT-
M, Poon W-S. Techniques for stereotactic
7. Herff C, Krusienski DJ, Kubben P. The potential 23. Jun JJ, Steinmetz NA, Siegle JH, et al. Fully inte- neurosurgery: beyond the frame, toward the
of stereotactic-EEG for brain-computer interfaces: grated silicon probes for high-density recording of intraoperative magnetic resonance imaging-
current progress and future directions. Front Neu- neural activity. Nature. 2017;551:232-236. guided and robot-assisted approaches. World
rosci. 2020;14:123. Neurosurg. 2018;116:77-87.
24. Goss-Varley M, Dona KR, McMahon JA, et al.
8. Chari A, Thornton RC, Tisdall MM, Scott RC. Microelectrode implantation in motor cortex 40. Jamjoom AAB, Jamjoom AMA, Marcus HJ.
Microelectrode recordings in human epilepsy: a causes fine motor deficit: Implications on poten- Exploring public opinion about liability and re-
case for clinical translation? Brain Commun. 2020;2. tial considerations to Brain Computer Interfacing sponsibility in surgical robotics. Nat Mach Intell.
fcaa082. and Human Augmentation. Sci Rep. 2017;7:1-12. 2020;2:194-196.

9. Budman E, Deeb W, Martinez-Ramirez D, et al. 25. Russell C, Roche AD, Chakrabarty S. Peripheral 41. Waziri A, Schevon CA, Cappell J, Emerson RG,
Potential indications for deep brain stimulation in nerve bionic interface: a review of electrodes. Int J McKhann GM, Goodman RR. Initial surgical
neurological disorders: an evolving field. Eur J Intell Robot Appl. 2019;3:11-18. experience with a dense cortical microarray in
Neurol. 2018;25, 434-e30. epileptic patients undergoing craniotomy for
26. Anderson JM, Rodriguez A, Chang DT. Foreign subdural electrode implantation. Neurosurgery.
10. Deep NL, Roland JT. Auditory brainstem im- body reaction to biomaterials. Semin Immunol. 2009;64:540-545 [discussion: 545].
plantation: candidacy evaluation, operative tech- 2008;20:86-100.
nique, and outcomes. Otolaryngol Clin North Am. 42. Bullard AJ, Hutchison BC, Lee J, Chestek CA,
2020;53:103-113. 27. Klopfleisch R, Jung F. The pathology of the Patil PG. Estimating risk for future intracranial,
foreign body reaction against biomaterials: fully implanted, modular neuroprosthetic sys-
11. Matias CM, Sharan A, Wu C. Responsive neuro- Foreign Body Reaction to Biomaterials. J Biomed tems: a systematic review of hardware complica-
stimulation for the treatment of epilepsy. Neurosurg Mater Res A. 2017;105:927-940. tions in clinical deep brain stimulation and
Clin North Am. 2019;30:231-242. experimental human intracortical arrays. Neuro-
28. Lotti F, Ranieri F, Vadalà G, Zollo L, Di Pino G. modulation. 2020;23:411-426.
12. Bouthour W, Mégevand P, Donoghue J, Invasive intraneural interfaces: foreign body re-
Lüscher C, Birbaumer N, Krack P. Biomarkers for action issues. Front Neurosci. 2017;11:497. 43. Seifried C, Weise L, Hartmann R, et al. Intra-
closed-loop deep brain stimulation in Parkinson operative microelectrode recording for the delin-
disease and beyond. Nat Rev Neurol. 2019;15: 29. Di Pino G, Formica D, Lonini L, et al. ODEs eation of subthalamic nucleus topography in
343-352. model of foreign body reaction around peripheral Parkinson’s disease. Brain Stimul. 2012;5:378-387.

146 www.SCIENCEDIRECT.com WORLD NEUROSURGERY, https://doi.org/10.1016/j.wneu.2020.11.028


LITERATURE REVIEW
ASWIN CHARI ET AL. BRAINeMACHINE INTERFACES: THE ROLE OF THE NEUROSURGEON

44. Lee PS, Weiner GM, Corson D, et al. Outcomes of and Information Processing (GlobalSIP). square-market-wisconsin-a7856971.html. Accessed
interventional-MRI versus microelectrode December 7e9, 2016; Washington, DC. https:// September 26, 2019.
recording-guided subthalamic deep brain stimu- doi.org/10.1109/GlobalSIP.2016.7906022
lation. Front Neurol. 2018;9:241.
62. Burwell S, Sample M, Racine E. Ethical aspects of
54. Pugh J, Pycroft L, Sandberg A, Aziz T, Savulescu J.
brain computer interfaces: a scoping review. BMC
45. da Silva NM, Ahmadi S-A, Tafula SN, et al. Brainjacking in deep brain stimulation and au-
Medical Ethics. 2017;18:60.
A diffusion-based connectivity map of the GPi for tonomy. Ethics Inf Technol. 2018;20:219-232.
optimised stereotactic targeting in DBS. Neuro-
image. 2017;144:83-91. _ HG. External relaying based
55. Kulaç S, Sazli MH, Ilk 63. Ienca M, Andorno R. Towards new human rights
security solutions for wireless implantable medical in the age of neuroscience and neurotechnology.
46. Opie NL, Oxley TJ. Removing the need for inva- devices: a review. Paper presented at: 2018 11th Life Sciences, Society and Policy. 2017;13:5.
sive brain surgery: the potential of stent elec- IFIP Wireless and Mobile Networking Conference
trodes. Bioelectronics in Medicine. 2019;2:9-11. (WMNC). September 3-5, 2018. Prague. https:// 64. Synchron: Team. Synchron. Available at: https://
doi.org/10.23919/WMNC.2018.8480911 www.synchronmed.com/team. Accessed November
47. Opie NL, John SE, Rind GS, et al. Focal stimula-
tion of the sheep motor cortex with a chronically 4, 2020.
56. Digital Health Center of Excellence. Artificial In-
implanted minimally invasive electrode array telligence and Machine Learning in Software as a
mounted on an endovascular stent. Nat Biomed Medical Device. FDA. Published online January 28, 65. NeuralinkeCrunchbase Company Profile & Fund-
Eng. 2018;2:907-914. 2020. Available at: https://www.fda.gov/medical- ing. Crunchbase. Available at: https://www.
devices/software-medical-device-samd/artificial- crunchbase.com/organization/neuralink. Accessed
48. Schevon CA, Ng SK, Cappell J, et al. Micro- August 31, 2020.
intelligence-and-machine-learning-software-medi
physiology of epileptiform activity in human
cal-device. Accessed April 7, 2020.
neocortex. J Clin Neurophysiol. 2008;25:321-330.
66. ScienceeHumm Helps You Think Better. Humm.
57. Zhang T, He Z, Lee RB. Privacy-preserving Ma- Available at: https://thinkhumm.com/science.
49. Rey HG, Pedreira C, Quian Quiroga R. Past,
chine Learning through Data Obfuscation. arXiv: Accessed September 5, 2019.
present and future of spike sorting techniques.
180701860 [cs]. Published online July 12, 2018.
Brain Res Bull. 2015;119:106-117.
Available at: http://arxiv.org/abs/1807.01860.
50. Frauscher B, von Ellenrieder N, Zelmann R, et al. Accessed April 7, 2020. Conflict of interest statement: A.C. is supported by a Great
Atlas of the normal intracranial electroencepha- Ormond Street Hospital Children’s Charity Surgeon Scientist
logram: neurophysiological awake activity in 58. Tully MP, Hassan L, Oswald M, Ainsworth J.
Fellowship. This work was supported by the National
different cortical areas. Brain. 2018;141:1130-1144. Commercial use of health dataea public “trial” by
Institute of Health ResearcheGreat Ormond Street Hospital
citizens’ jury. Learn Health Syst. 2019;3:e10200.
Biomedical Research Centre. HJM is supported by the
51. De Vico Fallani F, Bassett DS. Network neurosci-
59. Munyon CN. Neuroethics of non-primary brain Wellcome/EPSRC Centre for Interventional and Surgical
ence for optimizing brainecomputer interfaces.
computer interface: focus on potential military Sciences (WEISS) and NIHRBRC Neuro-Oncology Theme.
Phys Life Rev. 2019;31:304-309.
applications. Front Neurosci. 2018;12:696. Received 23 September 2020; accepted 5 November 2020
52. Bouchard KE, Aimone JB, Chun M, et al. High-
Citation: World Neurosurg. (2021) 146:140-147.
performance computing in neuroscience for data- 60. Drew L. The ethics of brainecomputer interfaces.
Nature. 2019;571:S19-S21. https://doi.org/10.1016/j.wneu.2020.11.028
driven discovery, integration, and dissemination.
Neuron. 2016;92:628-631. Journal homepage: www.journals.elsevier.com/world-
61. This tech company is becoming one of the first in neurosurgery
53. Hosseini M-P, Soltanian-Zadeh H, Elisevich K, the world to microchip employees. The Indepen-
Available online: www.sciencedirect.com
Pompili D. Cloud-based deep learning of big EEG dent. Published July 24, 2017. Available at: http://
data for epileptic seizure prediction. Paper pre- www.independent.co.uk/news/business/news/ 1878-8750/$ - see front matter ª 2020 Elsevier Inc. All
sented at: 2016 IEEE Global Conference on Signal us-tech-company-microchip-employees-first-three- rights reserved.

WORLD NEUROSURGERY 146: 140-147, FEBRUARY 2021 www.journals.elsevier.com/world-neurosurgery 147

You might also like