A new approach towards neuromodulation

A New Approach towards Neuromodulation
1. University of Western Ontario, London Health Sciences Centre 339 Windermere Road A10-026, London Health Sciences Centre
1. Introduction
Many neurological disorders are determined by degeneration of specific groups of cells. Parkinson's disease (PD) is
characterized by a loss of neurons from the substantia nigra which results in a deficiency of the neurochemical
levodopa. Replacement of dopamine by its precursor levodopa, in the long-term, results in motor fluctuations and
dyskinesias. From an electrophysiological point of view, the main hypothesis is that the subthalamic nucleus
becomes overactive changing the function of the internal segment of the globus pallidus (GPi) [1]. At the circuitry
level, these changes in subthalamic nucleus (STN) dynamically alter also decision thresholds which has profound
consequences regarding decision making [2-,3].
This clinical deterioration can be alleviated by a neuromodulatory surgical treatment. The literature is now well
established in showing that electrical modulation can alter the firing patterns of several nuclei upstream and close to
the site of stimulation. Deep brain stimulation (DBS) of subthalamic nucleus (STN) and of the pars interna of Globus
Pallidus (GPi) is an effective procedure able to control severe Parkinsonian symptoms [4].
However after DBS treatment, balance or gait disturbances can occur in some patients. Our hypothesis is that such
disturbances might occur as an effect of several factors. First, the complexity of neuronal circuitry involve the
cooperation of several nuclei including subthalamic-pallidal system The BG circuitry and its involvement in
generating complex motor tasks is not fully understood. Second, such local electrical stimulation may have collateral
effects in stimulating connected areas. Third, finding the right input, how to deliver stimulation, what patterns to
deliver in, among many other variables are factors that are extremely difficult to modulate. Additionally it is
important to understand the origin of complex electrical patterns as a result of changes in interactions between
nuclei. This outcome of alleviating previous problems and generating new issues is unexplained yet in
neurophysiological terms.
2. Neural Plasticity - From Recordings to Neuromodulation
There are three major issues in neuromodulation. The first issue is that of implantable technology. The second
problem is the dynamics of neuromodulation. The final and the most important concern is related to what we mean
by neuromodulation. We will briefly discuss each of these aspects in the next few paragraphs.
2.1 Implantable technology
The first issue is that of the technology that is being used for this purpose. It is well known that the current
technology is a modification of the cardiac pacemaker technology that is not really well suited or adapted for neural
application. The anatomical structure of the heart is relatively much simpler than even the most fundamental
neuronal structures. Since their introduction in the 1940’s, neuromodulation techniques have matured to the point
that neurosurgeons can record from, stimulate and produce lesions in deep brain structures. [5-6] and [7] used
microelectrodes in the early 1960’s to allow direct recording of single or multiunit activity in deep brain structures at
spatial intervals from microns and in a frequency range from 1-200 Hz . The recent resurgence of neuromodulation
procedures in deep brain structures has therefore been due not only to a better understanding of functional
neuroanatomy of cells involved in these diseases but also to the development of techniques for accurately localizing
these cells.
An implantable device comprising groups of microelectrodes bundled together at high-density (“multichannel
electrodes”) increases the resolution of individual passes, and can stimulate and record a 20-200 μm radius around an
inserted site. Stimulation with such an electrode produces current spread with a radius of up to 3mm and allows an
estimation of the proximity of the electrode to important motor, sensory or visual structures. Subsequent chronic high
frequency stimulation will reduce the activity of abnormal firing within the target neuron (U.S. Patent No. 1,662,446, issued to Wappler). Multi-channel electrodes, which combine the recording functions of microelectrodes and the stimulating functions of macro-electrodes, have been reported. Generally, these systems consist of recording and stimulating wires, which radiate from a planar backbone (see, e.g., U.S. Patent 5,282,468). Because of the large surface area these electrodes occupy, they generally are suited only for recording and stimulating neurons at the surface of the brain and are not for use in deep brain surgery. Deeper probes such as the Michigan probe have been used for surface recordings in animals. We are in the process of building a new technology that addresses many of the following important concerns. Our
electrode is unique. No such device is currently employed in neuromodulation. It has many biological and
engineering advantages which include improved target localization ability and reduced operating room time,
integrated chronic stimulation ability, and long term fine tuning of target eliminating repeat surgery.

2.2 Stimulation delivery

Despite these advances in technology, an important question that has really not been well answered, and indeed not
really asked, is what parameters are required for the optimal delivery of neuromodulation? The current approach is to
deliver an extremely standardized, uniform, non-physiological electrical impulse, as discussed above. Another
approach is to at some level, sense the intrinsic neuronal signal, and understand the dysfunction as in an epileptiform
discharge and then deliver impulses in response to this abnormal signal to somehow neutralize the signal. Finally,
new work that “reads” what the neural signal might be indicating in patients with disability is used to move cursors
and robotic arms to perform tasks. These various levels of alterations in the electrical properties of the neural signal
by delivery only or read-and-deliver paradigms have many limitations. Yet the one fundamental limitation is that we
do not have a good or even a basic understanding of the complexity of the underlying neural processing and
computation, before we can really comprehend the meaning of the term neuromodulation. The exact mechanism of
action of DBS is still unknown however, four hypotheses are proposed: a) an activation of inhibitory circuits within
the targeted nucleus [8] b) an antidromic stimulation of inhibitory afferents to the targeted nucleus concomitantly
with the release of GABA [9] c) a nonsynaptic blockade of neuronal transmission through the deactivation of
voltage-dependant ion channels [10] and d) a driving of efferent information and a jamming of the encoded one by a
superposition of a nonphysiologic high frequency pattern [11] However, if we are to really understand the
mechanisms of neuromodulation, the question really to ask is what indeed does one mean by neuromodulation? How
important is it to define this term at a scientific level? What impact can that have in the development of future
technologies? Is it enough to treat the neural system as an electrical entity and then deliver energy ad-hoc without
being able to conform to the intrinsic finesse that is built into it? Will our ability to perform neuromodulation ever
reach a level of complexity that would be necessary to truly modify the circuitry in such a way as to take advantage
of the vast computational power underneath it? To tackle these complex issues, a very new approach to
understanding neural function and computation has to be developed.
2.3 How far can neuromodulation be extended?

An overview of this approach will be provided next. We will explore concepts of information processing and flow.
The true meaning of neuromodulation will be framed within this context, and the suggestion of the necessary finesse
that is required to achieve this will be brought forth.

As a function of behavioral state neural circuits produce variable outputs [12-13] with neurotransmitter release
considered to be the key process for communication in the brain, a mechanism was firstly proposed in the fifties and
continuously reinforced lately by several electrophysiological observations. The popularity of extracellular
recordings in lab techniques has sustained a collective consciousness that the action potential is an invariant, all-or-
nothing stereotyped event [14-16] .Once the membrane reaches a certain threshold potential the invariant spike is
generated. This simplistic approach relates the spike frequency with the amount of transmitter release. Facilitation
and depression are considered to be the key neuromodulatory mechanisms. The signal carries information from the
neuronal cell body, the soma, down the axon to presynaptic terminals, where it evokes the release of neurotransmitter
to excite or inhibit the next neuron. Yet, it is unclear how these effects can alter neuronal outcome and highlight a
specific response. It is unlikely that this 'digital' electric signal carries full information about previous synaptic, soma
activations or the release of neurotransmitter and it is still unknown how transmitter release from spiking neurons in
brain is influenced by observed changes in membrane potential.
In vivo tetrode cell recordings from striatum show modulatory effects of electrical flow during AP occurrence during
action selection task at the critical point where action selection occurs [17-18]. Such alteration in electrical flow is
related to decision to follow a certain path on the T-maze. It is likely this electrical modulation is a result of memory
trace for synaptic events which took place in combination with the neuromodulator (dopamine) release. Even though
we are able to capture the modulation of electrical flow these recordings do not, however, provide continuous
monitoring of the fluctuations of membrane potential, and do not capture sub-threshold changes in membrane
potential such as those caused by individual synaptic events. Extracellular recordings cannot reveal several details
regarding cellular events which determine the intrinsic plasticity required for neuronal circuitry.
Recent electron microscopy studies showed how synaptic vesicles [19] release neurotransmitter etc. The exocytosis
of synaptic vesicles is mediated at the presynaptic active zone of nerve terminals. The trafficking cycle is triggered
by ionic mechanism (Ca2+ ions) which sustain rapid release . This macromolecular machine is protein regulated and
a thermodynamic approach is needed to understand intricacies of information transfer and computation [20]
However, the effects of dopamine on such neurons is complex, including increase of excitability or suppression of
certain paths for transmission.
Popular analyses in cognitive-neurophysiology projects have mainly paid attention to the quantification of firing
rates as markers of behavioral correlates of neurons. [12, 21]. A classical application of information theory computes
the mutual information. This is a measure that takes into account statistical dependency of recorded signals. Such
methods derived from information theory usually address the statistical question as to what extent a given neural
response codes the stimulus input. These analyses have been extended to neuronal ensembles and to relate the
activity in neural ensembles with changes in behavior. Therefore information transfer within neuronal ensemble and
changes in functional connectivity resulting from learning have already been explored [22].
Additionally, theoretical studies regarding information transfer using the Hodgkin-Huxley (HH) model showed that
there is a higher transfer of information provided by ionic fluxes than estimated by spiking theory approach [23].
This issue demonstrate the role of spikes in modulate information transfer at a level inaccessible for spike timing
approach. Such analysis extends the possibility of achieving new results in neuromodulation by targeting this
echelon within spikes alone.
We anticipate that the therapeutic effects of electrical modulation probably consist in its ability to change
pathological paths of electrical flow within targeted regions of the brain. We believe and have shown that in the
normal scenario, this electrical modulation is based upon the thermodynamic state and informational need within the
neuron and the network. Any neuromodulation process will eventually need to replicate these intricacies. To reliably
evoke small synaptic events, stimulation strengths should be set at extremely low values but be dynamic to be able to
actually reproduce the level of modulation that is occurring within the nervous system. By using the techniques of
neural computation and modeling the electromagnetic field we are able to understand the quantitative effects of
electrical modulation. Since slight changes in electrode position in brain or electrode geometry architecture can
strongly modify the consistency of electric field such electrical modulation has a determinant effect and consequent
neural response. New technologies must take these issues into account if true neuromodulation is to be provided.
3. Conclusion

The easiest way to investigate this process is to develop models and simulate on computer the electric field generated
by several shapes of electrodes. Another step requires the characterization of electrical properties of the brain tissue.
It is expected that the response of neuronal circuitry depends on the accuracy of the neuronal models. Since these
neurons have 3D geometries with several ion channels reconstructing their topology from experimental recordings is
an important step. However, current spike timing models [24, 14, 15, 25] are inadequate to sustain such level of
detail since biophysical properties of neurons where ion channel function has to be modeled in detail is important.
Such models require not only characterization of the activity of neurons in time domain they have to account also for
spatial properties of electrical modulation. New technologies will need to address these intrinsic complexities.
4. References
1 P. Limousin,Effect On Parkinsonian Signs And Symptoms Of Bilateral Subthalamic Nucleus Stimulation Lancet
345, 91 (1995)
2 M. J. Frank, Hold your horses: a dynamic computational role for the subthalamic nucleus in decision making.
Neural Netw. 19, 1120 , 2006 .
3 Kringelbach M.L., Jenkinson N., Owen S.L.F. & Aziz T.Z. (2007) Translational principles of deep brain
stimulation. Nature Reviews Neuroscience. 8:623-635. PMID 17637800.
4 Gildenberg Philip L. (2005) Evolution of neuromodulation. Stereotact Funct Neurosurg, 83(2-3), 71-79. PMID
16006778.
5.D . Albe-Fessard, G . Arfel, and G . Guiot., Activities electriques characteristiques de quelques
structures cerebrales ches l'homme. Ann Chir, 17: 1185--1214, 1963.
6. D . Albe-Fessard, G . Guiot, and J . Hardy. Electrophysiological localization and identification
of subcortical structures in man by recording spontaneous and evoked activities. Electroencephalogr Clin
Neurophysiol}, 15, 1963.
7 H . Jasper and G . Bertrand. Exploration of the human thalamus with microelectrodes. Physiologist, 7, 1963.
8 Strafella A, Ashby P, Munz M, Dostrovsky JO, Lozano AM, Lang AE. (1997). "Inhibition of voluntary activity by
thalamic stimulation in humans: relevance for the control of tremor." Movement Disorders 12: 727-737.
9 Dostrovsky JO, Levy R, Wu JP, Hutchison WD, Tasker RR, Lozano AM. (2000). "Microstimulation-induced
inhibition of neuronal firing in human globus pallidus." Journal of neurophysiology 84: 570-574.
10 Beurrier C, Bioulac B, Audin J, Hammond C. (2001). "High-frequency stimulation produces a transient blockade
of voltage-gated currents in subthalamic neurons." Journal of neurophysiology 85: 1351-1356.
11 Hashimoto T, Elder CM, Okun MS, Patrick SK, Vitek JL. (2003). "Stimulation of the subthalamic nucleus
changes the firing pattern of pallidal neurons." Journal of neuroscience 23: 1916-1923.
12 M.S. Jog, Y. Kubota, C.I. Connolly, Hillegaart, A.M., Graybiel, Building neural representations of habits, (1999)
Science 286, 1745-9.
13 M.S. Jog, C.I. Connolly, Y. Kubota, D.R. Iyengar, L. Garrido, R. Harlan, A.M. Graybiel, 2002, Tetrode
technology: advances in implantable hardware, neuroimaging, and data analysis techniques. Journal of Neuroscience
Methods 117,141-152
14 W., Maass, Networks of spiking neurons: the third generation of neural network models. Neural Networks, (1997)
10:1659-1671
15 W. Gerstner, and W.M., Kistler, Spiking Neuron Models Single Neurons, Populations, 2002, Plasticity
16 Rieke, F., Warland, D., de Ruyter van Steveninck, R., and Bialek, W. (1996). Spikes - Exploring the neural code.
MIT Press, Cambridge, MA.Cambridge University Press.
17 D Aur, MS Jog, Neuronal spatial learning, Neural Processing Letters, Vol 25, no 1, pp 31,47 2007a.
18 D Aur, and M Jog, Reading the Neural Code: What do Spikes Mean for Behavior?. Available from Nature
Precedings <http://dx.doi.org/10.1038/npre.2007.61.1, 2007b
19 Cowan, W.M., Sudhof, T.C. & Stevens, C.F. Synapses (Johns Hopkins University Press, Baltimore; 2000)
20 D Aur, and MS Jog. Beyond Spike Timing Theory – Thermodynamics of Neuronal Computation. Available from
Nature Precedings, http://hdl.handle.net/10101/npre.2007.1254.1 , 2007c
21TD Barnes, Y Kubota, DZ Jin, AM Graybiel Activity of striatal neurons reflects dynamic encoding and recoding
of procedural memories,Nature 2005 Oct 20 437(7062):1158-61
22 MS Jog, D., Aur, CI, Connolly Is there a Tipping Point in Neuronal Ensembles during Learning? Neuroscience
Letters Volume 412, Issue1 , 2007, Pages 39-44;
23 Aur D., Connolly C.I. and Jog M.S., (2006) Computing Information in Neuronal Spikes, Neural Processing
Letters, 23:183-199.
24 Bialek W., DeWeese M., Rieke F. and Warland D, (1993) Bits and brains: Information flow in the nervous
system Physica A: Statistical and Theoretical Physics, Volume 200, 1-4, 581-593
25 ZF Mainen and TJ Sejnowski Reliability of spike timing in neocortical neurons, Science, Vol 268, Issue 5216,
1503-1506,1996

Source: http://cnfrs.institut-telecom.fr/pages/pages_ursi/URSIGA08/papers/K04ap1.pdf