THEORETICAL AND EXPERIMENTAL PREDICTIONS OF NEURAL ELEMENTS
ACTIVATED BY DEEP BRAIN STIMULATION
by
SVJETLANA MIOCINOVIC
Submitted in partial fulfillment of the requirements
For the degree of Doctor of Philosophy
Dissertation Adviser: Dr. Cameron McIntyre
Department of Biomedical Engineering
CASE WESTERN RESERVE UNIVERSITY
August, 2007
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of
SVJETLANA MIOCINOVIC
candidate for the PhD degree*.
(signed) Dominique Durand (chair of the committee)
Ruth Siegel
Ken Gustafson
Jerrold Vitek
Cameron McIntyre
(date) 5/23/2007
* We also certify that written approval has been obtained for any proprietary material contained therein.
Dedicated to my parents
Posvećeno mojim roditeljima
TABLE OF CONTENTS
List of Tables ………………………………………………………………………….….2
List of Figures ……………………………………………………………………………..3
List of Abbreviations ……………………………………………………………………..5
Acknowledgements ……………………………………………………………………….7
Abstract …………………………………………………………………………………...9
Chapter 1: Introduction, background and project significance ………………………….11
Chapter 2: Software system for stereotactic neurosurgical navigation in non-human
primates (Cicerone)………….………………..….….….……………..…………40
Chapter 3: Spatial and temporal characteristics of voltage field generated by deep
brain stimulation electrode …………………….………….……………..………65
Chapter 4: Computational analysis of subthalamic nucleus and lenticular fasciculus
activation during therapeutic deep brain stimulation ……….……..…………….90
Chapter 5: Summary and future directions ….………………...….…………...….……124
Bibliography ……………………………………………………………………...……136
Appendix A: Cicerone user manual….…………………...….…………………………147
1 LIST OF TABLES
3-1. Parameters optimized for modeling DBS-induced voltage fields in saline and
subcortical gray matter – static model
3-2. Parameters optimized for modeling DBS-induced voltage fields in saline and
subcortical gray matter – Fourier model
2 LIST OF FIGURES
1-1. Deep brain stimulation
1-2. Simplified circuit diagram of the basal ganglia
1-3. DBS in a parkinsonian monkey
1-4. Neural environment in the subthalamic region
1-5. Spread of direct stimulation effects in a PD patient
2-1. Cicerone display and graphical user interface
2-2. Microelectrode recording data and Cicerone atlas
2-3. Cicerone atlas accuracy
3-1. In vitro voltage field recordings
3-2. Surgical planning and electrode implantation
3-3. Repeated in vivo DBS voltage measurements
3-4. In vivo voltage field recordings
3-5. Effect of DBS electrode impedance on recorded voltage
3-6. Stimulation waveforms from in vivo and in vitro recordings and corresponding
model solutions
4-1. Three-dimensional reconstruction of the basal ganglia
4-2. Multicompartment cable model of an STN projection neuron
4-3. Neural populations and DBS electrode in the context of 3D neuroanatomy
4-4. Field-neuron model of STN DBS
4-5. STN neuron firing in response to extracellular stimulation
4-6. Neural activation during clinically effective and ineffective DBS
3 4-7. Experimentally recorded GPi firing during STN DBS
4-8. Sensitivity of neural activation to electrode position
4-9. STN firing frequency under the influence of stimulation-induced trans-synaptic
GABAa inhibitory inputs
4 LIST OF ABBREVIATIONS
2D two-dimensional
3D three-dimensional
AC anterior commissure
AP anterior-posterior
CST corticospinal tract
CT computed tomography
DBS deep brain stimulation
EP entopeduncular nucleus
FEM finite element model fMRI functional magnetic resonance imaging
GABA gamma aminobutyric acid
GPe globus pallidus externus
GPi globus pallidus internus
HFS high frequency stimulation
IPG implantable pulse generator
L-dopa levodopa
LF lenticular fasciculus
MER microelectrode recording
ML medio-lateral
5 MPTP 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine
MRI magnetic resonance imaging
PD Parkinson's disease
PET positron emission tomography
SNc substantia nigra pars compacta
SNr substantia nigra pars reticulata
STN subthalamic nucleus
STN DBS subthalamic deep brain stimulation
STN HFS subthalamic high frequency stimulation
VD ventral-dorsal
VTA volume of tissue activated
VTK Visualization Toolkit
ZI zona incerta
6 ACKNOWLEDGEMENTS
I would first like to thank my advisor Dr. Cameron McIntyre. Cameron has been
an incredible source of support and guidance during my four years in his lab. He was
always able to put things into perspective and make me see what is truly important. He
was also always very encouraging of my professional development and sensitive to my
educational goals in the Medical Scientist Training Program.
Next, I would like to express gratitude to Dr. Jerry Vitek and his lab, who have
accepted me as one of their own. My biggest thanks goes to Dr. Gary Russo who has
taught me most of what I know about monkey electrophysiology. Drs. Weidong Xu and
Jianyu Zhang were always there to help and share their vast knowledge of experimental techniques. Jennie Minnich was an incredibly dedicated and capable lab manager. Her
predecessor Karen Zingale was always helpful as well, as were the wonderful people in
Biological Resources Unit who took care of the animals.
A big thanks goes to the members of my own lab. Scott Lempka was incredibly
patient and helped me do most of the experiments described in Chapter 3. Chris Butson
provided a lot of guidance in the modeling phase of the project. Angela Noecker offered
advice and technical assistance during Cicerone software development. Phil Hahn, Matt
Johnson, Chris Maks, and Luis Lujan contributed to the success of the project as well.
Everybody’s generosity and enthusiasm made our lab a wonderful and inspiring environment to work in.
My graduate school research experience started in the lab of Dr. Warren Grill who introduced me to the exciting field of neural engineering. I thank Dr. Grill for the early support and encouragement that set me on this path.
7 I would also like to thank Drs. Dominique Durand, Jerry Vitek, Ruth Siegel and
Ken Gustafson for having served on my thesis committee.
This research was financially supported by the National Institutes of Health (RO1
NS-47388, RO1 NS-37019, and T32 GM07250)
8 Theoretical and Experimental Predictions of Neural Elements Activated by Deep
Brain Stimulation
Abstract
by
SVJETLANA MIOCINOVIC
Chronic electrical stimulation of the brain, known as the deep brain stimulation
(DBS), has become the preferred surgical treatment for advanced Parkinson’s disease.
Despite its clinical success the mechanisms of DBS are still unknown and there is limited
understanding of the neural response to DBS. As a result the therapeutic neural target has
not been clearly identified, which limits opportunities to improve the technology and
increase treatment efficacy. We hypothesized that subthalamic (STN) projection neurons
are primarily activated during clinically effective STN DBS.
Non-human primate models of DBS provide unique opportunities to study the
therapeutic mechanisms of DBS in vivo. The therapeutic benefits of DBS are dependent
on accurate placement of the electrode in the appropriate neuroanatomical target.
Stereotactic neurosurgical navigation systems that exist for clinical applications are lacking in the area of non-human primate research. Therefore, we developed a software system (Cicerone) for stereotactic neurosurgical planning, neurophysiological data collection, and DBS visualization in primates.
Computational volume conductor models are commonly used to estimate neuronal response to electrical stimulation. To date there has been no direct validation of
9 models aimed at investigating stimulation of subcortical structures. We have therefore
measured voltages generated by DBS electrode in the thalamus of a monkey.
Furthermore, we have calculated model parameters that can be used to accurately capture
both spatial and temporal properties of voltage fields induced by DBS.
Utilizing the stereotactic navigation system and voltage field model we built a
comprehensive computational model of STN DBS in the parkinsonian monkey. We
compared our model predictions with results from experimental animals to quantify the
relative activation of STN neurons and pallidothalamic (GPi) fibers during therapeutic
DBS. The results indicate that activation of nearly half of the STN neurons is sufficient
for the behavioral manifestation of the therapeutic effects, which confirms our hypothesis. The additional recruitment of GPi fibers of passage may also play an important role in therapeutic outcome, but large-scale activation of GPi fibers is not necessary. The position of the electrode in the STN region and the choice of active contact can strongly effect recruitment of either neural population.
10 Chapter 1: Introduction, background and project significance
1.1 Deep brain stimulation in treatment of Parkinson’s disease
Deep brain stimulation (DBS) has become a preferred surgical therapy in the
treatment of movement disorders (Benabid et al., 1996; Obeso et al., 2001) and it is being
investigated as a treatment for epilepsy (Hodaie et al., 2002), obsessive-compulsive
disorder (Gabriels et al., 2003) and depression (Mayberg et al., 2005). It has been
particularly successful for patients in the late stages of Parkinson’s disease (PD) when
medications are no longer sufficient to alleviate the disease symptoms. Tremor, rigidity
and bradykinesia (slowness of movement) are the main manifestations of PD and initially
the drug levodopa can ameliorate most symptoms. However after 5-10 years of levodopa
therapy, 50-75% of patients experience two major side-effects: motor fluctuations
(sudden return of parkinsonian state) and drug induced dyskinesias (involuntary, often
painful movements of arms, legs and head including abnormal postures) (Lewis et al.,
2003). These symptoms are the major indication for DBS surgery (Lozano and Mahant,
2004).
In DBS, high frequency electrical stimulation is applied through a four-contact
electrode placed permanently in the brain (Fig. 1-1). For PD the preferred location for
electrode placement is the subthalamic nucleus (STN), although globus pallidus internus
(GPi) is also an effective target (Vitek JL, 2002a). The electrode is connected to a
battery-powered pulse generator implanted in the patient’s chest. The pulse generator produces monopolar, pseudo-monophasic cathodic, rectangular voltage pulses with amplitude usually between 1-3.5 V, pulse duration between 60-210 µs and frequency
11
Figure 1-1. Deep brain stimulation. Electrode is implanted permanently in the motor control region of the brain. Electrical pulse generator is placed underneath the skin and connected to the electrode, providing constant stimulation.
range from 130-185 Hz. After the implantation surgery the patient undergoes extensive clinical testing to determine, by trial and error, which set of parameters produce the best reduction of symptoms while minimizing side effects.
In many patients, DBS successfully alleviates many disabling symptoms of PD; however, in others it has only partial benefit or it produces side effects which limit usefulness of the stimulation therapy. Studies of the clinical outcome of STN or GPi DBS have shown improvements of 41% to 67% in standardized ratings of motor symptoms
(Greenberg and Rezai, 2003) and 43% improvement in the overall quality of life
(Benabid, 2003). These numbers reflect significant clinical improvements, but sometimes fall short of the patient’s expectations (Ford et al., 2004). In addition, a number of
12 adverse effects can be generated by DBS including sensorimotor impairments,
involuntary movements (stimulation-induced dyskinesias), as well as speech, mood and
cognitive disturbances (Volkmann et al. 2002; Krack et al., 2002; Benabid 2003; Lozano
and Mahant 2004; Okun et al. 2005). Often these side-effects can be avoided or alleviated
with proper adjustment of the stimulation settings (Krack et al. 2003). However, further
improvements in the engineering design and clinical implementation of DBS technology will rely on addressing a number of questions on the effects of DBS on the nervous system. The fundamental goal of this thesis was to enhance our understanding of the target neural elements of the stimulation.
1.2 Current understanding of DBS mechanisms
Deep brain stimulation (DBS) is currently the main surgical therapy for treatment of advanced Parkinson’s disease (PD). It has largely replaced ablative surgery because it is reversible and adjustable, permitting titration of stimulation parameters in an attempt to
optimize therapeutic benefit with minimal side effects. The DBS surgery can be
performed bilaterally without the high incidence of side effects associated with bilateral ablative surgery. Furthermore, the use of DBS technology does not preclude patients from possibly benefiting from future advances in PD treatment. Despite its remarkable clinical success, the therapeutic mechanisms of DBS are still not completely understood, limiting opportunities to improve treatment efficacy and simplify selection of stimulation parameters. The three main questions are: 1) how does DBS affect individual neurons
13 around the electrode; 2) what is the neural target of therapeutic DBS; and 3) how does
DBS influence the cortico-thalamo-basal ganglia network?
1.2.1 Current controversies: inhibition vs. excitation
Since the inception of DBS as a clinical therapy, its mechanisms have been the
focus of intense scientific study. The similarity between the clinical effects of DBS
(Benazzouz et al., 1993; Limousin et al., 1995) and those resulting from a lesion in the
same nucleus (Bergman et al., 1990; Aziz et al., 1991; Patel et al., 2003) prompted the
initial theory that DBS works by inhibiting neural output (Dostrovsky et al., 2000;
Beurrier et al., 2001). These conclusions were based on single unit extracellular
microrecordings from cell bodies that showed high frequency stimulation resulted in
reduction of action potential firing in the stimulated nucleus. However recent studies
suggest a more complex mechanism of action that includes stimulation-induced
activation of axons. Observations from a variety of experimental studies demonstrated
changes in neuronal activity that were consistent with activation of output from the
stimulated structure (Hashimoto et al., 2003; Vitek 2002; Windels et al., 2000, 2003; Jech
et al., 2001; Perlmutter et al., 2002). Although these studies seemed in direct
contradiction to earlier reports of inhibition at the stimulated site, modeling studies
suggested that neurons could be inhibited at the stimulated site, while axons projecting
from the inhibited neurons and axon passing nearby the electrode would be activated by
stimulation (McIntyre et al., 2004). Combined analysis of these electrophysiological,
biochemical, functional imaging, and modeling data sets indicated that neural activation was a fundamental component of DBS. This in turn, led to more intense study of which
14 neural elements are directly stimulated by DBS and how is activation dependent on the stimulation parameter settings.
It now appears that in addition to the direct stimulation of STN projection neurons, stimulation of fiber tracts and other nuclei surrounding the STN may also be responsible for the therapeutic outcome associated with DBS in the subthalamic region.
Further, the direct stimulation effects of DBS spread throughout the basal ganglia thalamocortical network. As a result of these findings, recent hypotheses on the therapeutic mechanisms of DBS have concentrated on stimulation induced regularization or interruption of pathologic neural activity to explain the reduction of disease symptoms.
1.2.2 Methodologies for studying DBS mechanisms
A variety of research modalities, each with their own strengths and weaknesses, have been employed to study mechanisms of DBS. Electrophysiological recordings of neuronal activity, biochemical analyses, computer modeling and imaging studies have all provided crucial pieces to the mechanism puzzle. The goal is to integrate the complementing and contrasting results from these methods to develop a comprehensive understanding of DBS mechanisms.
Neural recording studies have included experiments performed on brain slices, anesthetized or awake animals, and PD patients undergoing DBS surgery. In vitro brain slice preparations make it possible to record neural activity intracellularly and characterize cell membrane properties in detail (Beurrier et al., 2001; Magarinos-Ascone et al., 2002; Garcia et al., 2003, 2005; Lee et al., 2003, 2006). Various agonists and antagonists can easily be added to the slices to further characterize neural response.
15 Similarly, stimulation can be applied to the tissue to simulate DBS and determine the
effects of stimulation on neuronal elements in the slice. Stimulation amplitude and
current spread in a thin slice, however, may not accurately reflect the in vivo conditions.
Unlike in vitro experiments, in vivo studies have the advantage of preserving brain structure and anatomical connections and provide the ability to correlate neuronal activity to behavior. In vivo experiments record extracellular activity and their primary disadvantage is the presence of stimulus artifact due to recording amplifier saturation during stimulation. If the artifact is not removed, it can obscure activity during the first few milliseconds after a stimulus pulse. In some studies due to a large artifact neural activity is recorded immediately following DBS. However post-stimulation activity may not accurately reflect neural response that occurs during stimulation. Methods to remove or reduce this artifact have been developed and they have significantly prolonged the
duration of usable signal (Hashimoto et al., 2002). In vivo microelectrode recordings are
done primarily in human PD patients (Dostrovsky et al, 2000; Wu et al., 2001 ; Pralong et
al., 2003; Filali et al, 2004; Welter et al., 2004; Galati et al., 2006; Montgomery, 2006),
monkeys (Boraud et al., 1996; Hashimoto et al., 2003;Anderson et al., 2003; Bar-Gad et
al., 2004; Meissner et al., 2005; Kita et al., 2005) and rats (Benazzouz et al., 1995, 2000;
Tai et al., 2003; Maurice et al., 2003; Lee et al., 2004; Shi et al., 2006). Human PD
patients, while naturally ideal subjects, can only participate in microelectrode recordings
during medically warranted procedures due to the significant risk from and invasiveness
of deep brain recordings. As a result, most human microrecordings are done during DBS
implantation surgery in the same nucleus where the DBS electrode is implanted. The
DBS electrode itself can also be used to record local field potentials (LFPs) in the
16 targeted nucleus (Brown et al., 2001; Marsden et al., 2001; Foffani et al., 2003, 2006). As an alternative, monkeys can be used to carry out microrecordings in nuclei other than the implanted nucleus and provide invaluable information on the downstream effects of DBS.
Similarly, rats can be used to record at multiple sites within the basal ganglia. Rat and monkey models of PD are now commonly utilized in DBS studies.
Biochemical studies of DBS mechanisms include experiments that measure levels of various neurotransmitters, secondary messengers and mRNA in specific nuclei throughout the basal ganglia. These studies are typically performed in rats, although there have been several reports of neurotransmitter measurements from human PD patients
(Stefani et al., 2005, 2006; Galati et al., 2006). A microdialysis technique is utilized to measure levels of glutamate, GABA, dopamine, or secondary messengers (e.g. cGMP) by extracting small fluid samples from targeted region in either an anesthetized or awake subject (Windels et al., 2000, 2003, 2005; Meissner et al., 2002, 2003, 2004; Bruet et al.,
2003; Boulet et al., 2006). The advantage of this method over neural recordings is that there is no interference from electrical stimulus artifact during DBS, but temporal resolution is much poorer (samples are generally collected every 15 minutes). To achieve real-time monitoring, constant current amperometry can be employed to monitor efflux of neurotransmitter such as dopamine (Lee et al., 2006). In this case, however, the presence of dopamine is only indirectly demonstrated using its electrochemical properties.
Neurotransmitter analysis has been used to help establish the mechanism underlying neuronal responses during stimulation e.g. it can provide support for activation of certain pathways by finding increased levels of the neurotransmitter used by that pathway.
However, the source of neurotransmitter cannot be determined with certainty, although
17 synaptic origin is commonly assumed. In situ hybridization is another approach that has been used to study DBS mechanisms. This technique can investigate molecular changes in the basal ganglia, but requires that the animal is euthanized. Changes in neuronal metabolic activity can be inferred by measuring levels of cytochrome oxidase subunit I
(CoI) mRNA (Salin et al., 2002; Benazzouz et al., 2004). Expression of neurotransmitter- related genes (e.g. GAD67, substance P and enkephalin) and immediate early gene encoded proteins (e.g. c-fos) can be used to monitor cellular response to various experimental conditions (Salin et al., 2002; Bacci et al., 2004; Schulte et al., 2006). In this way short or long term adaptive mechanisms occurring in the brain in response to
DBS can be studied.
Computer modeling studies simulate experiments in a highly controlled environment. Model neurons and neural circuits can be perturbed at will and their responses observed at the level of a network, single neuron, or ion channel. The goal of modeling studies is to explain experimental findings by reducing the complexity of the system and identifying those variables responsible for the observed phenomena. In addition, computer models can generate viable hypotheses that can be tested experimentally. Anatomically and biophysically realistic models of basal ganglia neurons and their interaction with detailed representations of DBS electric fields are currently available (McIntyre et al., 2004; Miocinovic et al., 2006; Butson et al., 2007). These field-neuron models can be used to predict the response of each neuronal element (soma, axon, dendrites) to stimulation. Comprehensive subject-specific DBS models can be created by building a 3D representation of the subject’s basal ganglia and reconstructing the precise DBS electrode position within the target site. The nuclei are then populated
18 with model neurons and the calculated electric field is applied to individual cells. The
response of model neurons to DBS can then be correlated to the observed clinical effect
in that particular patient or experimental animal. However, the electric field generated by
a complex geometry electrode and its effect on neurons has not been validated in vivo which has been a major shortcoming in DBS modeling studies.
Imaging studies are a powerful way to examine the effect of DBS at the network level by simultaneously observing activity at multiple sites in the brain. The most popular
studies use positron emission tomography (PET) and functional magnetic resonance
imaging (fMRI) to observe changes in brain circuitry on the basis of changes in regional
blood flow and blood oxygenation, respectively (Ceballos-Baumman, 2003). It is assumed that these increases or decreases in neuronal activity stem from changes in afferent synaptic activity or local interneuron activity, rather than from changes in output
from the imaged region (Perlmutter et al., 2002). In addition to functional imaging, cortical activity during DBS can be studied with electroencephalographic (EEG), electromyographic (EMG) and transcranial magnetic stimulation (TMS) techniques
(Ashby et al., 1999, 2001; Marsden et al., 2001; Baker et al., 2002).
1.2.3 Effect of DBS on neuronal activity
The nuclei of the basal ganglia are linked into a complex network through both inhibitory and excitatory connections (Fig. 1-2). The input into this network comes mainly from the cerebral cortex and the thalamus, whereas the output is directed toward the thalamus and the brainstem areas, primarily the pedunculopontine nucleus (PPN) and midbrain extrapyramidal area (MEA). A stimulating electrode positioned in the brain
19 exerts both local and distal effects. Local effects can be investigated by observing the
neuronal response in the stimulated nucleus. Distal effects can be studied by examining
the effect of DBS on sites downstream and upstream of the stimulated nucleus as the
stimulation response spreads antidromically and orthodromically throughout the basal
ganglia-thalamo-cortical network. Elucidation of the mechanisms of DBS requires
understanding of both local and distal effects of the stimulation.
Figure 1-2. Simplified circuit diagram of the basal ganglia (Alexander and Crutcher, 1990). The loss of dopaminergic output from SNc in PD causes a cascade of alterations affecting the entire circuit. Relative line thickness indicates degree of activation; arrows represent excitatory and circles inhibitory connections; blue represents GABAergic, red glutamatergic and green dopaminergic neurons. STN = subthalamic nucleus; GPe = globus pallidus externus; GPi = globus pallidus internus; SNr = substantia nigra pars reticulata; SNc = substantia nigra pars compacta. Coronal brain section from Martin and Bowden (2000).
1.2.4 Neuronal recordings in the stimulated nucleus
The STN is positioned at a central location within the basal ganglia and has direct influence over the major output structures of the basal ganglia, GPi and substantia nigra pars reticulata (SNr). STN neurons are spontaneously active due in part to the pacemaker
20 activity of persistent sodium channels (Bevan and Wilson, 1999). This activity is further modulated by excitatory synaptic connections from the cerebral cortex and to a lesser extent the thalamus, and inhibitory afferents from the GPe (Hamani et al., 2004). As a result, STN neurons fire at a frequency of about 20 Hz (Wichmann et al., 1994). In PD they become hyperactive and spike in a bursty and irregular manner, with an average firing rate of ~40 Hz (Bergman et al., 1994; Hutchison et al., 1998; Magnin et al., 2000;
Benazzouz et al., 2002). This hyperactivity is thought to increase the inhibitory drive of the basal ganglia on the thalamus and suppress thalamic excitatory output, resulting in reduced cortical activity and the appearance of hypokinetic motor symptoms (DeLong,
1990). Based on this model of PD, lesions in the STN or GPi would reduce the excessive inhibition of the GPi and improve the motor signs associated with PD. In support of this hypothesis lesions in either the STN or GPi are associated with improvement in PD motor signs. The earliest hypotheses on DBS mechanisms attempted to reconcile the similarity in clinical outcome after a lesion and during DBS by proposing that high frequency stimulation inhibited neurons and decreased output from the stimulated site.
Numerous studies have shown that high frequency stimulation in the STN suppresses activity of STN neurons. In a recent study, Meissner et al. (2005) recorded
STN neuronal activity for several minutes before, during, and after high frequency
stimulation (HFS) with microelectrodes that improved contralateral rigidity in parkinsonian monkeys (100µA amplitude, 130 Hz frequency and 60µs pulse width).
Removal of the stimulus artifact by a template subtraction method allowed neural activity
to be recorded during stimulation with minimal loss of the recorded signal. They showed
that therapeutic stimulation decreased the mean firing rate in the majority of STN
21 neurons, from 19 Hz to 8 Hz. Activity returned to baseline within 100 ms following the
end of the stimulus train. Meissner et al. (2005) proposed that the decrease in the mean firing rate resulted from the resetting of the firing probability of STN neurons to virtually zero by each stimulus pulse. Neurons resumed activity after about 3 ms following a stimulus pulse and returned to baseline firing probability after approximately 7 ms.
Stimulation at 130 Hz corresponds to 7.7 ms interpulse interval, meaning that the neurons were able to fire at their baseline rate for only a very short period of time, resulting in an overall reduction in the mean firing rate.
Further evidence in support of somatic inhibition in the stimulated nucleus during
DBS comes from recordings in the STN and GPi. The studies of STN DBS in humans showed that neuronal activity in almost all STN cells examined was reduced or completely inhibited during stimulation (Filali et al., 2004; Welter et al., 2004). About
50% of the cells were completely inhibited for approximately 150 ms following a
stimulus train while the remaining cells showed no consistent effect (Filali et al., 2004).
The reduction in neuronal activity was observed only at frequencies above 40Hz, exhibiting similar frequency dependence as the therapeutic DBS effects (Welter et al.,
2004). Studies in rats have also shown that STN activity is inhibited during therapeutic
STN DBS in both awake subjects (Shi et al., 2006) and anesthetized animals (Tai et al.,
2003; Benazzouz et al., 2000). It should also be noted that similar results were observed in pallidal neurons during GPi HFS in both human patients (Dostrovsky et al., 2000; Wu et al., 2001) and parkinsonian monkeys (Boraud et al. 1996; Bar-Gad et al., 2004).
The above mentioned studies demonstrate that HFS reduces somatic activity of local neurons, so the question arises as to what causes this apparent inhibition and more
22 importantly how relevant is this local observation in explaining the therapeutic effects of stimulation. Since electrical stimulation is generally thought to excite neurons, mechanisms proposed to explain the inhibition are depolarization block - cessation of activity due to increase in membrane potential and inactivation of sodium channels
(Beurrier et al., 2001; Benazzouz et al., 1995), and synaptic inhibition - stimulation- induced activation of inhibitory presynaptic terminals (Dostrovsky and Lozano, 2002).
Support for the depolarization block hypothesis comes mainly from in vitro experiments.
STN cells have been shown to increase firing during the initial stimulation period after which they fail to respond suggesting inactivation of sodium channels (Magarinos-
Accone et al., 2002; Lee et al., 2006). But another in vitro study found that STN HFS can generate spike bursts time-locked to stimulus pulses (Garcia et al., 2003). In vivo studies favor the synaptic inhibition hypothesis, and the fact that in vitro slices are disconnected from their inhibitory inputs could explain the different results observed in the two types of studies. In an in vivo situation, depolarization block is unlikely because STN HFS reduces but does not completely block neuronal activity (Tai et al., 2003; Welter et al.,
2004; Meissner et al., 2005), inhibition can occur even after a single stimulus pulse (Filali et al.; 2004), and both inhibition and recovery from inhibition occur at latencies consistent with GABAergic postsynaptic current kinetics (Meissner et al., 2005).
Furthermore, in vitro STN HFS can either excite or inhibit STN neurons through a synaptically mediated mechanism (Lee et al., 2003, 2004). Indeed, a small number of
STN neurons have also been excited in vivo, which could result from activation of excitatory afferents which are also present in the STN (Tai et al., 2003). Stimulation induced synaptic inhibition could also explain inhibitory effects seen during GPi HFS
23 since the GPi also receives strong inhibitory connections from GPe. Interestingly, one study on thalamic DBS has shown that thalamic neurons which receive predominantly excitatory afferents can be excited by stimulation (Dostrovsky and Lozano, 2002).
1.2.5 Neural recordings in downstream nuclei
In the previous section evidence was presented that DBS inhibits somatic activity in the stimulated nucleus. However, inhibition of somatic activity does not necessarily reflect reduced output from the nucleus. Indeed, several studies have suggested that output is increased from the stimulated nucleus (Hashimoto et al., 2003; Anderson et al.,
2003; Maurice et al., 2003) bringing into question the mechanism underlying this dissociation. One explanation is that axons are excited, while the cell soma is suppressed.
Axons are the neural elements most easily excited by extracellular stimulation, and they are likely to be activated by DBS. It is exceptionally difficult to directly record axonal activity; however, axonal firing can be indirectly recorded by monitoring the post- synaptic cell. STN neurons send excitatory glutamatergic projections to the GPe and the two output structures of the basal ganglia, GPi and SNr. Recordings in these target nuclei reflect downstream effects of DBS which are of crucial importance to understanding mechanisms of DBS.
Hashimoto et al. (2003) demonstrated that neuronal activity in GPe and GPi increased in response to STN DBS suggesting increased output from the STN (Fig. 1-3).
The experimental setup in awake, parkinsonian monkeys closely resembled human DBS system. Monkeys were implanted with a scaled down version of a clinical DBS electrode
(4 contacts with 0.75 mm diameter, 0.5 mm height and 0.5 mm separation between the
24
Figure 1-3. DBS in a parkinsonian monkey. A. A monkey was made parkinsonian by administration of MPTP. DBS electrode was implanted in the STN and internal pulse generator was placed in the back. Stimulation was applied through the DBS electrode while microelectrode recordings were performed in the GPe and GPi. Adapted from Elder et al., 2005. B and C. Examples of neuronal responses occurring during STN stimulation in a GPi and GPe cell, respectively. Top traces show analog signal overlays of 100 sweeps made by triggering at 10 msec intervals in the prestimulation (before start of stimulation) period and by triggering on the stimulation pulse in the on-stimulation period. Arrows indicate residual stimulation artifacts after artifact template subtraction. Middle traces display peristimulus time histograms (PSTHs) reconstructed from successive 7.0 msec time periods in the prestimulation period and from the interstimulus periods, in the on-stimulation period. The first PSTH bin is omitted in the on-stimulation period because of signal saturation and residual stimulation artifacts. *Significant increase at p<0.01; † significant decrease at p<0.01; Wilcoxon signed rank test. Bottom plots represent the mean firing rate calculated every 1 sec on the basis of the PSTH illustrating the time course of the firing rate. From Hashimoto et al., 2003.
25 contacts) and an implantable pulse generator. Therapeutic stimulation parameters resulted
in reduction of contralateral rigidity and bradykinesia and an increase in spontaneous
movement. Microelectrode recordings were performed in GPe and GPi during both effective and ineffective stimulation, and peri-stimulus time histograms were constructed.
During therapeutically effective stimulation, a majority of neurons showed a significant increase in the average firing rate. In addition there was a consistent pattern of response of these neurons to STN stimulation with two consistent peaks of increased activity in the post-stimulus time histogram occurring at 3 ms and 6.5 ms. Surrounding the excitatory peaks were periods of inhibition, especially pronounced in GPi neurons, which is not surprising since they receive GABAergic connections from the GPe. The precise pattern and latency of the responses resulted in regularization of GPe and GPi activity. During therapeutically ineffective stimulation, the overall firing rate did not change significantly.
These results suggest that therapeutic STN DBS activates subthalamopallidal projections and changes the discharge pattern of GP neurons from irregular to a stimulus- synchronized, more regular pattern of activity, which the authors hypothesized was responsible for the reduction in parkinsonian symptoms. Similar excitatory latencies in
GPe and GPi neurons during STN stimulation in monkeys were observed by Kita et al.
(2005). This study also showed that the excitatory response in GPe neurons was glutamatergic while GPi inhibition was GABAergic and originating from GPe. The inhibitory GPi response seen in this study was more pronounced than those observed in the Hashimoto et al. (2003) study.. The relative importance of inhibitory GPe-GPi connections compared to excitatory STN-GPi connections in non-parkinsonian animals stimulated with small electrodes and long pulses could contribute to the observed
26 differences. In further support of the ‘output activation’ hypothesis in non-human
primates, Anderson et al. (2003) showed that GPi stimulation decreased thalamic activity
consistent with activation of inhibitory GABAergic GPi projections to thalamic neurons.
Recordings in nuclei receiving STN inputs in rats generally support the notion
that STN output is activated by DBS, although some results vary across studies. Shi et al.
(2006) simultaneously recorded at multiple locations within the basal ganglia and found
nearly equal numbers of excitatory and inhibitory responses in GP (rat analog of primate
GPe) and SNr during therapeutic STN DBS which improved treadmill locomotion in
parkinsonian rats (50-175 µA, 130 Hz, 60 µs, intermittent 3-s on, 2-s off cycle).
Similarly, in normal, anesthetized rats, Maurice et al. (2003) showed that STN DBS
causes inhibition of SNr neurons at low amplitudes (20-80 µA), but excitation at higher
amplitudes (100-240 µA). The authors suggested that inhibitory effects were likely due to activation of inhibitory pallidonigral fibers or GABAergic intranigral neurons, whereas excitatory effects resulted from direct activation of subthalamonigral projections. Several studies also in normal, anesthetized rats show the opposite effect, a decrease in SNr firing during high amplitude STN DBS (400 µA; Tai et al., 2003) and long-lasting inhibition in
SNr and entopeduncular nucleus (EP; rat analog of primate GPi) immediately following high amplitude STN DBS (300-500 µA; Benazzouz et al., 1995, 2000). However,
Benazzouz et al. (2000) also found a long-lasting excitation in GPe which would be consistent with activation of excitatory STN output. Differences in stimulation amplitudes, exact stimulation sites and anesthesia methods likely contribute to the different results seen in these studies. These differences however, also attest to the complex pattern of excitation and inhibition that is likely to emerge in response to
27 stimulation and the importance of considering the effects of stimulation on polysynaptic
pathways (e.g. STN-GPe-SNr) and fiber tracts surrounding the stimulation site (e.g.
nigrostriatal, pallidothalamic, pallidonigral, and cerebellothalamic in the case of STN
DBS).
Several studies in human PD patients where recordings in downstream nuclei
were possible also support the output activation hypothesis (Pralong et al., 2003; Galati et
al., 2006; Montgomery et al., 2006). Galati et al. (2006) recorded an increase in firing
frequency and more regularity in the firing pattern of SNr neurons during therapeutic
STN DBS. Montgomery (2006) reported a reduction in thalamic neuronal activity 3.5-5
ms following a stimulus pulse during GPi DBS, consistent with orthodromic inhibition of
thalamic neurons by activated GPi axons. In addition, human PET studies showed an
increase in blood flow in the region of GPi during STN DBS (Hershey et al., 2003) and
an increase in cortical blood flow during thalamic DBS both consistent with activation of
output from the stimulated site (Perlmutter et al., 2002). Similarly, an fMRI study found
an increase in blood oxygen level-dependent signal in the GPi of patients undergoing
STN DBS (Jech et al., 2001).
Even though cell bodies in the stimulated nucleus are inhibited, the axons of these
projection neurons seem to be activated. This causes downstream excitation, when
glutamatergic STN neuron axons are activated; inhibition, when GABAergic GPi neuron
axons are activated; or a combination of excitation and inhibition, when polysynaptic
pathways are involved. For example, the GPi response to STN DBS is influenced by
direct excitatory STN-GPi projections and indirect inhibitory STN-GPe-GPi pathway.
The role of antidromic axonal activation should also be considered, such as the activation
28 of afferent cortical projections during STN DBS which may affect cortical and subsequently striatal activity (Ashby et al., 2001; Kita et al., 2005). In summary, neural recording studies suggest that DBS inhibits local cell bodies likely by activating inhibitory presynaptic terminals. At the same time it also activates projection axons of the local neurons and surrounding fibers of passage resulting in a complex pattern of excitatory and inhibitory effects which modulate not only local basal ganglia activity but the basal ganglia thalamocortical network as a whole.
1.2.4 Effect of DBS on the cortico-thalamic-basal ganglia network
The classical model of the basal ganglia (Alexander and Crutcher, 1990) predicts that destruction of nigrostriatal dopaminergic neurons as seen in PD leads to hyperactivity of the STN. This causes excessive activity in the GPi and subsequent inhibition of thalamocortical projections thought to result in the motor signs associated with PD. Both STN and GPi lesioning reduce the hyperactivity in the STN-GPi circuit removing the excessive thalamo-cortical inhibition and resulting in alleviation of parkinsonian symptoms. Although this model can explain the improvement in parkinsonian motor signs seen with lesions in the STN and GPi, it does not explain the improvement in dyskinesia seen with GPi lesion or the improvement in some parkinsonian motor signs following thalamotomy. This has led to the hypothesis that patterns of neuronal activity may be more important than changes in rate in the development of parkinsonian motor signs. Consistent with this hypothesis STN DBS increases STN output and the mean discharge rate of GPi neurons, yet similarly leads to
29 improvement in motor function. The key to this apparent paradox is the stimulation frequency necessary to achieve therapeutic effect.
A proposed mechanism of DBS that is consistent with an increase in neural output from the STN is that stimulation overrides pathological neuronal discharge by imposing a more regular higher frequency neuronal activity from the STN to the GPi (Montgomery and Baker, 2000; Vitek, 2002). Both experimental (Hashimoto et al., 2003; Garcia et al.,
2005) and modeling (Grill et al., 2004; McIntyre et al., 2004) studies have shown that high frequency stimulation replaces intrinsic irregular activity with one that is time- locked to the stimulus. Only frequencies above 100 Hz provide symptom relief while frequencies below 20 Hz often worsen symptoms probably because this just adds spikes to an already irregular pattern of spontaneous firing. Neurochemical studies have also shown that low frequency stimulation does not lead to the molecular changes seen with high frequency stimulation (Windels et al., 2003; Schulte et al., 2006). It has been suggested that regularization of GPi firing by STN DBS restores the responsiveness of thalamocortical cells to synaptic inputs (e.g. sensorimotor information) despite increased inhibitory drive (Rubin and Terman, 2004).
Since an action potential in response to extracellular stimulation initiates in the axon, high enough stimulation frequency can override neurons intrinsic output by two possible mechanisms. First, antidromic stimulus-initiated action potentials can collide with orthodromic soma-initiated spikes blocking this irregular pattern of activity from being conducted down the axon and second, antidromic invasion of the soma prevents the cell from discharging spontaneously due to the refractory period associated with such activity. In both cases irregular orthodromic activity is replaced by a more regular pattern
30 of discharge. Even though this tonic, high frequency firing pattern is not considered
normal, it is seemingly devoid of informational content and results in an ‘informational lesion’, preventing the pathological activity from being transmitted within the basal ganglia (Grill et al., 2004). Interestingly, in dystonia, where intrinsic pathologic GPi firing rates are lower than in PD, therapeutic DBS frequencies may also be lower
(Tagliati et al., 2004).
The analysis of the current DBS experimental data supports the concept that the neural pattern, rather than firing rate, is an important determinant of the pathologic state of PD and the therapeutic effects seen with DBS (Montgomery and Baker, 2000; Vitek,
2002; Hashimoto et al., 2003). In addition to an increase in the mean rate and irregularity of neuronal discharge in the basal ganglia, PD is also characterized by the development of rhythmic, oscillatory pathological activity (Bergman et al., 1994; Magnin et al., 2000).
Most notably, synchronized bursting is present between STN and GPe (Plenz and Kitai,
1999; Magill et al, 2001; Brown, 2001), and STN oscillatory frequency in the 15-30 Hz
(beta) range tends to predominate (Levy et al., 2002). Similar to the effect of L-dopa,
STN DBS appears to suppress abnormal beta rhythms in the GPi (Brown et al., 2004), but it is unclear if reduction of beta activity is necessary for symptom improvement (Foffani et al., 2006). STN DBS has been shown to decrease oscillatory and burst activity in the
STN and its target nuclei (Hashimoto et al., 2003; Meissner et al., 2005; Shi et al., 2006).
As a result, reduction of pathologic activity and its transmission, and the subsequent improvement in information processing could be responsible for amelioration of motor symptoms during DBS.
31 1.3 Therapeutic target of subthalamic deep brain stimulation
When considering therapeutic mechanisms of DBS, the primary focus has been on
the response of the neurons in the stimulated site, i.e. target nucleus neurons. However,
stimulation effects can and do spread outside the borders of the anatomical nuclei. This
finding is especially true for the STN, which is a small nucleus surrounded by several
major fibers tracts (Fig. 1-4; Hamani et al., 2004). Electrodes placed in the STN are
surrounded by local cells (neurons whose cell bodies are close to the electrode), afferent
fibers (neurons whose axonal terminals are near the electrode, but whose cell bodies are far away) and fibers of passage (neurons whose axons are passing near the electrode, but both cell bodies and axonal terminals are far away from the electrode). There is only
limited knowledge of the response properties of these different neural types to DBS.
The STN is an ovoid-shaped nucleus, ~10mm in size along the long axis in
humans and ~3.5mm in macaques (Hardman et al., 2002). STN projection neurons send
their axons to the globus pallidus, striatum and substantia nigra (Sato et al., 2000) while
the nucleus receives input from the external segment of the globus pallidus and the
cerebral cortex (Carpenter et al., 1981; Nambu et al., 1996). The lateral border of the
STN is defined by the internal capsule and the corticospinal tract (CST) fibers running through it. Lenticular fasciculus (LF) runs dorsal to the STN carrying inhibitory fibers
from the GPi to motor thalamus (Parent and Parent, 2004). The LF and the zona incerta
(ZI), which is a small nucleus dorsal to the LF, have been implicated as possible
32
Figure 1-4. Neural environment in the subthalamic region. DBS electrode is surrounded by STN projection neurons (left), GPi fibers of passage (lenticular fasciculus or Forel field H2) (middle) and corticospinal tract fibers (right). Afferent fibers not shown (see text). Each leg of 3D scale bar is 1 mm (D = dorsal, P = posterior, M = medial). Diameters of neuronal processes were increased ten times for figure rendering.
therapeutic targets because of their proximity to active DBS electrode contacts. In addition to pallidothalamic fibers several other fiber tracks pass in close proximity to the
STN, such as reciprocal STN-GPe projections, STN-substantia nigra pars reticulata connections, and nigrostriatal fibers. Neural recordings and biochemical studies in rats have suggested that pallidonigral and nigrostriatal fiber tracts are likely activated during
STN DBS and may contribute to its therapeutic effects.
The DBS electrode has four contacts and in each patient it is determined empirically which contact provides maximum relief of symptoms with minimum side effects. Intraoperative neurophysiological data, brain atlases and post operative imaging can be used to determine the location of the active contact with respect to the target site and the surrounding structures. The accuracy of these methods remains to be determined, but they can give an approximate idea of the location of the active contact. Numerous studies have supported the hypothesis that the optimal contact for therapeutic stimulation is located near the dorsal border of the STN, where stimulation effects are likely to
33 extend into the LF and ZI (Starr et al., 2002; Lanotte et al., 2002; Saint-Cyr et al., 2002;
Voges et al., 2002; Hamel et al., 2003; Yelnik et al., 2003; Herzog et al., 2004;
Zonenshayn et al., 2004; Nowinski et al., 2005; Yokoyama et al., 2006; Godinho et al.,
2006). The dorsolateral portion of the STN projects to cortical motor areas and is considered the “motor” area of the nucleus. As such one would argue that stimulation in this region would produce the greatest therapeutic benefit for motor symptoms. However, the current likely spreads outside the motor area and affects passing fiber tracts, including the internal capsule and the corticospinal tract. It is not clear if stimulation of the structures dorsal to STN is more therapeutic than stimulation of the STN itself, or if possibly the combined stimulation of multiple structures nearby make the STN such a successful target for PD treatment. Several studies found that stimulation dorsal to the
STN was less effective than stimulation at the dorsal border of the nucleus (Herzog et al.,
2004; Yokoyama et al., 2006). However, another study comparing stimulation in the
STN, at the dorsal border of the STN, and in the caudal ZI found that stimulation in the
ZI was the most effective (Plaha et al., 2006). Support for the argument that stimulation in ZI may be equally or more effective than in the STN is derived from reports of the involvement of ZI in locomotor activity and the observation that stimulation in the ZI in rats normalized dopamine-depletion induced molecular changes in a manner similar to that reported with STN stimulation (Perier et al., 2000; Benazzouz et al., 2004). Despite the intriguing results, the major deficiency of these studies is the inaccuracy in contact location estimation and inability to quantify the effective current spread.
Previous studies have implied, based on the presumed location of the active contact, that structures adjacent to the site of stimulation are also activated, however it is
34
Figure 1-5. Spread of direct stimulation effects in a PD patient. A and B. Volume of tissue activated (VTA) within STN, thalamus, and ZI/LF for contact 1 and contact 2 as a function of the stimulus voltage. Also shown are example VTAs at −2 V and −4 V. Relief of rigidity was observed at -2 V and 4 V for both contacts, while bradykinesia was improved at -2 V, but worsened at -4 V. From Butson et al., 2007.
not known precisely how far the stimulation spreads and how this varies with different stimulation parameters (amplitude, pulse width, frequency). A recent computer modeling study quantified the spread of stimulation during STN DBS in a human PD patient and correlated it to clinical outcomes (Fig. 1-5; Butson et al., 2007). The location of electrode contacts was reconstructed from postoperative MRI, and a 3D brain atlas was warped to the patient’s MRI to identify anatomical structures and their location with respect to the electrode. The patient-specific volume of tissue activated (VTA) was constructed using theoretical models of the DBS voltage field and neural (axonal) response to extracellular stimulation. The patient was clinically evaluated for rigidity, bradykinesia and
35 corticospinal tract activation at various stimulation parameter settings. The VTAs
accurately predicted the spread of stimulation into the internal capsule for stimulus
parameters that activated corticospinal fibers as measured by electromyography. Two separate contacts provided relief of rigidity and bradykinesia, and both of their VTAs included the LF and ZI. Improvement of rigidity was also correlated with spread of stimulation into the thalamus, the ZI, and the LF, but not the STN (contact closest to the
STN induced side effects so it was clinically ineffective). While this study analyzed a single patient, its conclusions agree with previous clinical studies and suggest that the ZI
and LF are activated during STN DBS and may provide a significant contribution to the beneficial effects observed with STN stimulation. The exact contribution of each, however, remains unclear.
1.4. Electrical stimulation in the central nervous system
The electric field generated by DBS is a complex phenomenon that is distributed throughout the brain (McIntyre et al., 2004b). This field is applied to the complex three- dimensional geometry of the surrounding neural processes (i.e. axons and dendrites). The response of an individual neuron to the applied field is related to the second derivative of the extracellular potential distribution along each process (McNeal, 1976; Rattay, 1986).
Each neuron surrounding the electrode will be subject to both depolarizing and hyperpolarizing effects from the stimulation (McIntyre and Grill, 1999; Lee et al., 2003).
As a result, in response to extracellular stimulation a neuron can be either activated or
suppressed in different neural processes depending on its position with respect to the
36 electrode and the stimulation parameters (McIntyre and Grill, 2000; McIntyre and Grill,
2002; McIntyre et al., 2004a).
Both local cells and fibers of passage respond at similar extracellular stimulation
thresholds (Ranck, 1975; Gustafsson and Jankowska, 1976; Nowak and Bullier 1998a,b;
McIntyre and Grill, 2000) making it difficult to determine which neurons are activated during STN DBS. Attempts have been made to use temporal excitation properties of
neurons, such as the chronaxie time and the refractory period, to infer which neural types
respond to DBS. Chronaxie time (Tch) is defined as the shortest duration of an effective
electrical stimulus with an amplitude equal to twice the minimum amplitude needed for
excitation with a pulse of infinite duration. Measurements of short chronaxie times and
refractory periods were used to conclude that large diameter axons were responsible for
DBS effects in subthalamic nucleus (Ashby et al., 1999, 2001; Hutchinson et al., 2002)
and globus pallidus (Ashby et al., 1998; Holsheimer et al., 2000; Wu et al., 2001).
However, both modeling (Miocinovic and Grill, 2004) and experimental studies (Nowak
and Bullier, 1998a; Swadlow, 1992) have shown that for extracellular stimulation
temporal excitation properties do not vary significantly between cells and axons. This
follows from the observation that during extracellular stimulation of local cells action
potential initiates in the axon and not in the cell body so the chronaxie time and refractory
period depend primarily on the axonal excitation characteristics. Measuring neuronal
temporal excitation properties is therefore not an adequate method for determining DBS target elements.
Neurophysiological studies show that STN neurons are inhibited by DBS; however, the output from the nucleus is increased. While it might seem improbable that a
37 neuron can be inhibited and activated at the same time during extracellular electrical stimulation; both experimental (Nowak and Bullier, 1998) and computer modeling
(McIntyre and Grill 1999; McIntyre et al., 2004) studies support this concept. The key finding that explains this apparent paradox is that when a cell is exposed to extracellular stimulation, the stimulation-induced action potential initiates in the axon rather than the cell body. In that case, inhibition of the cell body by whatever mechanism will not interfere with the neuron’s axonal output. A modeling study of thalamocortical neurons stimulated by DBS found that the position of the neuron with respect to the electrode determines the neuron’s output firing pattern (McIntyre et al., 2004). A neuron close to the stimulating electrode will have its spontaneous activity suppressed by activation of inhibitory presynaptic terminals, but its axon will be directly activated. As a result, the neuron will generate spikes time-locked to the stimulus frequency. A neuron positioned further away from the electrode will still be influenced by inhibitory synapses because axonal terminals are the most excitable neural elements (Baldissera et al., 1972;
Gustafsson and Jankowska, 1976). However, the stimulus will be subthreshold for direct axonal activation and the neural output will resemble that of the cell body, which can be a total or partial inhibition.
1.5. Project significance
DBS is the preferred surgical treatment for late-stage PD and it holds promise as a therapy for a number of other nervous system disorders. It is presently unknown what neural elements are responsible for therapeutic benefits in DBS, and this limits our ability to understand and optimally utilize this technology. The goal of this thesis was to identify
38 the neural element(s) responsible for the therapeutic effects of STN DBS. To address this goal we used detailed computational models of DBS in parkinsonian monkeys. An integral part of the model was a representation of the electric field generated by the DBS because field gradients are responsible for neural excitation. We hypothesized that the spatial and temporal characteristics of DBS voltage field can be accurately modeled using numerical finite element techniques. After describing the electric field generated by DBS, the next step was to evaluate the neural response to DBS. We hypothesized that STN projection neurons are primarily activated during therapeutic STN DBS.
The significance of this project was threefold. First, a neurosurgical navigation and visualization tool for studying DBS in non human primates was developed (Chapter
2). This software has become an integral part of animal workup in our lab, and it has been shared with several other labs working in the field of monkey neurophysiology. Second, voltage field models were validated by directly measuring potentials generated in the tissue by DBS electrode (Chapter 3). The in vivo data was used to estimate parameters for realistic modeling of DBS-induced voltage fields. Third, neural elements activated by therapeutic stimulation were identified by performing detailed computational analysis of
STN DBS in parkinsonian monkeys (Chapter 4). Together these achievements will improve accuracy and validity of both experimental and computational studies of DBS.
Furthermore, the results highlight the importance of considering stimulation spread outside the targeted anatomical areas.
39 Chapter 2: Software system for stereotactic neurosurgical navigation in non-human primates
Miocinovic S, Zhang J, Xu W, Russo GS, Vitek JL, McIntyre CC. Stereotactic neurosurgical planning, recording, and visualization for deep brain stimulation in non- human primates. J Neurosci Methods. 2007 May 15;162(1-2):32-41.
2.1. Introduction
Non-human primate models of deep brain stimulation (DBS) for Parkinson’s disease provide unique opportunities to study the therapeutic mechanisms of DBS in vivo
(Hashimoto et al., 2003; Elder et al., 2005). The therapeutic benefits of DBS are dependent on accurate placement of the electrode in the appropriate neuroanatomical target. As a result, stereotactic neurosurgical navigation systems for human (clinical) applications continue to evolve (Finnis et al., 2003; D’Haese et al., 2005). However, similar image-guided systems for non-human primate research are lacking. Therefore, we developed computerized techniques to address several limitations in traditional non- human primate stereotactic neurosurgery and microelectrode recording (MER).
The first limitation in non-human primate stereotactic surgery is the use of a standard 2D brain atlas for identification of the initial anatomical target location for DBS electrode implantation (or any similar procedure). This approach is prone to error because the brain size, shape, and location of subcortical structures can vary between animals
(Percheron and Lacourly, 1973; Francois et al., 1996; Deogaonkar et al., 2005). There have been numerous attempts to refine non-human primate neurosurgery with the
40 integration of magnetic resonance imaging (MRI) and population-based brain atlases
(Saunders et al., 1990; Alvarez-Rojo et al., 1991; Rebert et al., 1991; Nahm et al., 1994;
Asahi et al., 2003; Frey 2004; Francois et al., 1996; Deogaonkar et al., 2005; Christensen
et al., 1997). Unfortunately, most of these advances have not achieved wide scale
acceptance into practice.
Limitations also exist in the techniques used to collect and analyze MER data.
MER is performed to confirm and further explore the target brain region. However,
multiple sources of error exist such as brain shift, electrode deflection, improper electrode
zeroing and microdrive imprecision. These issues can cause discrepancy between the
expected electrode location in the brain and the experimenter defined classification of the
recorded neural signal. Brain atlases are commonly employed to provide anatomical
guidance in interpreting (classifying) the MER data because of variations in neural firing
characteristics. Unfortunately, traditional 2D brain atlases only provide information in the
sagittal or coronal plane, limiting opportunities to determine the electrode location along
oblique trajectories. In addition, MER data are typically saved as paper records, limiting options for data visualization.
Another limitation in current DBS practices is the lack of opportunity to visualize the predicted spread of stimulation for a given DBS electrode location in the brain. The fundamental purpose of DBS is to modulate neural activity with applied electric fields.
Therefore the ability to predict the spread of stimulation, prior to permanent implantation,
would provide additional information for identification of the optimal electrode placement in the brain. And, following implantation, predicting the spread of stimulation
41 could make the identification of therapeutic stimulation parameters a more
straightforward process.
To address these limitations we developed a software system (Cicerone) for stereotactic neurosurgical planning, neurophysiological data collection, and DBS
visualization. The system was developed for monkey DBS studies, but it can be applied
to a range of procedures requiring stereotactic localization and neurophysiological
mapping of subcortical structures in non-human primates. Cicerone provides interactive
3D visualization of co-registered MRI/CT images, subject-specific 3D anatomical brain atlas, and neurophysiological MER data. The software can be used to define a pre- operative target location and trajectory for the DBS electrode placement and help select the location on the skull for chamber placement. Entering microdrive coordinates and
MER data during recording sessions enables real-time interactive visualization of the electrode location in the brain. Cicerone also provides tools to compensate for stereotactic inaccuracies, thereby improving the match between the MER data and the brain atlas defined neuroanatomy. In addition, the user can simultaneously visualize the
DBS electrode and its predicted stimulating effects on the surrounding neural tissue. As a result, stereotactic placement of the DBS electrode can be optimized prior to permanent implantation using the combination of anatomical, neurophysiological, and electrical data. The goal of this study was to demonstrate the utility of Cicerone system for DBS electrode implantation in two rhesus macaques, targeting the subthalamic nucleus (STN) in one animal and the globus pallidus (GP) in the other.
42 2.2 Methods
The Cicerone software was written using VTK (Visualization Toolkit; Kitware,
Clifton Park, NY) and Tcl/Tk (Tool Command Language; http://tcl.sourceforge.net) making it portable across platforms, including Windows.
2.2.1 Subject-specific 3D brain atlas
Cicerone uses a 3D brain atlas that can be modified to fit the neuroanatomical profile of a specific animal. The standard atlas was created from the University of
Washington digital brain template atlas of the longtailed macaque (Macaca fascicularis)
(Martin and Bowden, 2000). This was accomplished by outlining individual structures in
each coronal atlas slice, spaced at 1 mm intervals. The 3D volumes were created by
interpolating between these contour lines using the graphical modeling program
Rhinoceros v3.0 (McNeal & Associates, Seattle, WA). With appropriate modifications
this atlas can be used for several other non-human primate species, including rhesus
macaque (Macaca mulatta) (see Discussion).
In this study we addressed two methods (linear scaling and nonlinear warping) for
adapting the standard 3D brain atlas to a specific subject. The simple approach was to
scale the standard 3D atlas along three axes, mediolateral (ML), anteroposterior (AP) and ventrodorsal (VD), to match the subject’s MRI and/or CT scans. This was accomplished in Cicerone with a simple graphical user interface that allows for manual scaling and repositioning of individual nuclei to better fit imaging and/or electrophysiological data.
The landmarks we used to create the linearly-scaled custom atlas were the anterior (AC) and posterior (PC) commissures, lateral ventricles and ML, AP and VD extent of the
43 brain, but other landmarks visible in the MRI could be used as well. The more intensive,
subject specific 3D atlas was constructed by warping 2D digitized brain atlas templates to
the corresponding MRI slices using Edgewarp (Bookstein, 1990; Martin and Bowden,
2000). Edgewarp applied a nonlinear warping function to atlas templates based on
manual landmark selection. When generating the warped atlas, we used the same
landmarks listed above, as well as the borders of visible nuclei such as the caudate,
putamen, globus pallidus, thalamus, and the optic tract. The warped atlas slices were then
converted into 3D volumes as described above using Rhinoceros v3.0 (Fig. 2-1 A).
The standard atlas currently contains more than twenty individual nuclei and
structures. To improve visualization, Cicerone only displays nuclei selected by the user and provides variable opacity and color for each structure.
2.2.2 Cicerone Setup
Imaging data (MRI and CT) are used to customize the 3D brain atlas to the specific animal and to set up a reference coordinate system. In our two monkeys, a CT was acquired to visualize the skull and the external landmarks (ear canals and orbital ridges) so that the stereotactic reference frame could be registered with the internal brain structures. An MRI was acquired to customize the 3D atlas and align the CT with the atlas. The MRI and CT were co-registered in Analyze 6.0 (AnalyzeDirect, Lenexa, KS) using an intensity-based mutual information algorithm (Viola and Wells, 1997). The co- registered imaging data was then imported into Cicerone as VTK volume files.
44
Figure 2-1. Cicerone display and graphical user interface. (A) A customized 3D atlas is aligned with the animal’s MRI. (B) The skull contour is extracted from the CT data and co-registered with the MRI and 3D brain atlas. Head chambers are interactively positioned within the stereotactic space defined by ear bars and orbital bars. (C) Theoretical volume of tissue activated by the DBS electrode is displayed within the STN of monkey 04-m-001 for the given set of stimulation parameters (5V, 135 Hz, 90µs, bipolar stimulation). (D) The user interface provides functions for manipulating the imaging data, 3D brain atlas, chambers, electrodes and defining MER locations.
45 Cicerone’s default coordinate system is based on the AC-PC plane. The center of
the AC is defined as the origin and the horizontal plane is perpendicular to the vertical plane through the AC-PC line. Although in monkey stereotactic neurosurgery it is
common to use a frame and an atlas referenced to the orbitomeatal plane. This is a plane
defined by the interaural line (line between tips of the earbars) and the inferior orbital ridges. The origin is defined as the midpoint of the interaural line and the orbitomeatal plane (Frankfurt zero). In Cicerone, the user can position earbars and orbital ridge bars to
define the orbitomeatal plane using the skull rendering extracted as a contour from the
CT data (Fig. 2-1B). The user can then switch from the ‘AC-PC coordinate system’ to the
‘stereotactic coordinate system’ (i.e. orbitomeatal plane). When moving to the
‘stereotactic coordinate system’, the origin is shifted to Frankfurt zero and rotation around ML axis is performed so that the orbitomeatal plane is horizontal.
If a subject-specific atlas has not been previously created, the user can scale the brain atlas in Cicerone to fit the imaging data. All files (MRI, CT, brain atlas), user preferences, and transformations performed during the set-up are stored in a configuration file for easy upload during later sessions.
2.2.3. Stereotactic neurosurgical planning
Cicerone provides tools to interactively position up to three recording chambers on the animal’s skull. This makes it possible to ensure, prior to surgery, that the electrodes can reach the desired target areas and that the chambers and electrode trajectories will not interfere with each other once the chambers are placed on the skull.
The chambers can be positioned in any oblique plane, providing an advantage over
46 currently used methods which allow only coronal or sagittal orientation as dictated by
traditional 2D print atlases. Chambers of any size and shape can be integrated into
Cicerone. Two chamber designs were used in this study. Both were cylindrical with 19.1 mm and 15.1 mm inner diameters, respectively. The user can also choose between several angles on the base of the chamber to find one that best fits the contour of the skull in a chosen location prior to surgery. The chamber coordinates provided by Cicerone can be applied directly to the stereotactic frame, after the frame is properly zeroed.
Cicerone’s current algorithm for coordinate calculation is designed for use with the Kopf frame (model 1430 and electrode manipulator model 1460; David Kopf Instruments,
Tujunga, CA). The coordinates of the recording site markers in 3D space (i.e. electrode tip location) are determined by constructing a transformation matrix based on the rotation angles and translation distances of the head chamber and electrode. The order of transformation operations is set so that the movement of virtual chamber/electrode reflects physical movement of the electrode manipulator on the stereotactic frame and the microdrive (i.e. the electrode manipulator sets the chamber position on the skull while the microdrive sets the electrode position within the chamber). The resulting matrix:
M = T2*Ry1*Rx2*T1*Rx1, where Rx1 is -90 degree rotation around ML axis necessary because x and y axes are reversed in Rhinoceros and VTK (Rhinoceros is used to build the head chambers and electrodes which are then imported into Cicerone); T1 is a translation matrix incorporating chamber movement along the ML and VD axes and electrode movement within the chamber along all three axes (the order of these translation coordinates depends on the orientation of the physical electrode manipulator since there are two
47 possible orientations on the frame); Rx2 is rotation around the ML axis set by the vertical
dial of the electrode manipulator; Ry1 is rotation around the VD axis set by the horizontal
dial of the electrode manipulator; and T2 is translation set by the chamber AP movement.
After the chamber placement surgery it is advisable to perform another CT scan to verify the chamber position on the skull.
2.2.4 Neurophysiological data collection and visualization
During MER data collection the user can view the electrode position within the
3D anatomical space by entering microdrive coordinates into Cicerone (Fig. 2-1D). The microelectrode recording (or stimulation) locations are displayed as small markers. The marker color is set according to their likely anatomical location (based on neuronal firing properties) and the marker shape is determined based on the presence or absence of a sensorimotor response. The user has the option to translate or scale the 3D atlas nuclei to better match the MER data as well as individually adjust the depth of each recording track to compensate for improperly zeroed electrodes. The user can pan, zoom and rotate the view, adjust the display properties of the 3D atlas, and scroll through the MRI in three orthogonal planes. The anatomy and the site markers can also be viewed in 2D, one slice at a time, which is sometimes preferable to a 3D view. The MER data are saved to a text file and can be re-imported into the program at a later time or stored to a central database.
2.2.5 Theoretical DBS volume of activation
A novel feature of Cicerone is its ability to predict the volume of tissue activated
(VTA) by a DBS electrode for a given electrode position and stimulation parameter
48 setting (Fig. 2-1C). Currently, Cicerone contains a database of pre-calculated VTAs for a
4 contact cylindrical monkey DBS electrode (0.75 mm diameter; 0.5 mm contact height).
Our methods of VTA calculation have been previously described (Butson and McIntyre,
2005; 2006; Butson et al., 2006a,b; Miocinovic et al., 2006). Briefly, the volumes were created by calculating activation thresholds for multi-compartment cable models of straight myelinated axons (2 µm in diameter) built in NEURON simulation environment
(Hines and Carnevale 1997). More than a hundred axons were placed perpendicular to an axisymmetric finite element model (FEM) of the cylindrical monkey DBS electrode in a
6x6 mm grid and stimulated with extracellular voltages generated in the tissue medium by the electrode. Activation thresholds (DBS voltage necessary to produce a propagating action potential in the axon models) were calculated for each axon and the results were used to extract contours representing VTAs for stimuli of various amplitudes (1-5 V).
The same procedure was repeated for different electrode configurations (monopolar or bipolar), stimulation pulse widths (60-210 µs), and frequencies (100-185 Hz). The VTAs do not vary based on the electrode location because the bulk tissue conductivity was assumed to be homogenous and isotropic for VTA calculation.
2.2.6 Chamber implantation and microelectrode mapping
Chamber placement, microelectrode mapping and DBS electrode implantation were performed on two female rhesus monkeys (Macaca mulatta; ID numbers 04-m-005 and 04-m-001) weighing 5.2 kg and 6.7 kg, respectively. All surgical and behavioral protocols were approved by the Institutional Animal Care and Use Committee and
49 complied with United States Public Health Service policy on the humane care and use of
laboratory animals.
Imaging. T2-weighted MRI scans were performed using a Siemens MAGNETOM
Trio 3 Tesla scanner (Siemens Medical Systems, Iselin, NJ) with the monkeys under propofol anesthesia. Twenty 2 mm thick coronal slices and twenty 2 mm thick sagittal slices were imaged (256x256 pixels; 0.47x0.47 mm in plane resolution). The two volumes were co-registered and fused using Analyze 6.0 software (AnalyzeDirect,
Lenexa, KS). CT scans were performed with a Siemens SOMATOM Sensation. CT
images were acquired in the axial plane in 0.6 mm or 1 mm increments (135 slices at
512x512; 0.24x0.24 mm in plane resolution). The CT volume was co-registered with the
fused MRI images in Analyze.
Surgical Procedures. A hemiparkinsonian syndrome was induced by unilateral
intracarotid injection of the neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine
(MPTP, 0.5 – 0.6 mg/kg over the course of ~15 minutes) during aseptic surgical
procedure under isoflurane anesthesia. After a stable hemiparkinsonian state was
achieved, craniotomies were trephined and recording chambers were implanted over the
craniotomy in a subsequent aseptic procedure under isoflurane anesthesia with the head
held in a primate stereotactic frame (Kopf frame model 1430). In each animal, two
craniotomies were performed and two metal chambers were anchored over the right
cerebral hemisphere. The chambers’ stereotactic coordinates were determined using
Cicerone as described in section 2.3.
In monkey 04-m-005, the chamber to be used to later implant the DBS electrode in the globus pallidus (GP) was placed in the coronal plane at 30 deg from the midline
50 (lateral-to-medial), 12.5 mm anterior and 2.1 mm lateral (in frame coordinates). The DBS electrode was implanted in a separate procedure after several weeks of mapping during which electrophysiological data was collected through this chamber. The second recording chamber to be used for later microelectrode recording in the thalamus, was placed in the sagittal plane at 20 deg from the vertical axis (anterior-to-posterior), 6 mm anterior and 5.5 mm lateral (in frame coordinates). Postoperative CT indicated that the location of the GP DBS chamber was 0.2 mm more lateral and 0.7 mm more anterior than planned, and the location of the thalamic recording chamber was 0.3 mm more medial and 0.4 mm more posterior than planned. These discrepancies were due to inherent inaccuracies in the surgical procedure, such as minor head misalignment in the stereotaxic frame and head movement due to forces exerted during craniotomy.
In monkey 04-m-001, the chamber later to be used to implant the DBS electrode in the STN was placed at an oblique orientation so that the DBS electrode would be oriented along the long axis of the nucleus. The chamber was angled 30 deg from the vertical axis and 15 deg from the sagittal plane (lateral-to-medial), 5 mm lateral and 11.2 mm anterior (in frame coordinates). The second chamber to be used for later microelectrode recording in the thalamus was placed in the sagittal plane at 25 deg
(posterior-to-anterior), 8.4 mm lateral and 23.2 mm anterior (in frame coordinates). Post- operative CT indicated that both chambers were misplaced by a five degree angle which was consistent with an unintentional head tilt during surgery due to an improperly positioned palate bar. Prior to starting microelectrode mapping the post-operative CT was imported into Cicerone and the virtual chambers were repositioned to reflect their actual positions on the skull.
51 Microelectrode mapping. Extracellular neuronal activity was recorded though the
chambers targeting GP and STN, respectively, to find the optimal position for the DBS
electrode implantation. Epoxylite-coated tungsten microelectrodes (0.5-1.0 MΩ) were
positioned within the chamber using a microdrive (MO-95-lp, Narishige Scientific
Instruments, Tokyo, Japan). Because the recording chambers were stereotactically positioned on the skull, the frame was not used for MER mapping or the subsequent DBS electrode implantation. During MER and DBS electrode implantations animals were sitting in primate chairs with their heads stabilized with a metal head post. Recording penetrations were made in coronal planes in monkey 04-m-005 and in oblique planes (15
degrees from sagittal plane) in monkey 04-m-001. The analog neuronal signal was
amplified (A-M systems model 3000, Sequim, WA), bandpass filtered at 500-3000Hz
(Krohn-Hite model 3380, Brockton, MA), digitally sampled at 25kHz and stored for later
offline analysis (Power 1401 and Spike2 software, Cambridge Electronic Design,
Cambridge, UK). Based on their discharge patterns together with other physiological
features and anatomical landmarks, neurons were classified as being located in the
striatum, globus pallidus externus (GPe), globus pallidus internus (GPi), STN or
unclassified. Neurons in the striatum exhibited typical low tonic firing rates and
characteristic injury discharge in response to electrode movement (Delong, 1971).
Transition into the pallidum was characterized by increased background activity and high
tonic firing rates (Turner and Anderson, 1997). The border between GPe and GPi was
characterized by little or no single neuron activity as the electrode tip crossed the internal
medullary lamina and by the presence of “border neurons” characterized by a more tonic
and lower frequency discharge pattern. In addition, most GPe neurons could be
52 characterized as either ‘pausers’ with high frequency discharge rates separated by long
pauses, or ‘bursters’ with low mean discharge rates and periodic bursts. GPi neurons
were more likely to fire continuously at high frequencies without long periods of silence
(DeLong, 1971). The STN was recognized by a dramatic increase in background activity as the electrode exited the internal capsule and by single neuron or multiunit activity that often discharged in characteristic bursts (DeLong et al., 1985). The optic tract was identified by the change in baseline noise level of neural recordings in response to flashes from a strobe light (Photic stimulator model PS33, Grass-Telefactor, Astro-Med, Inc.,
West Warwick, RI) in a darkened room. GPe, GPi and STN neurons were also tested for sensorimotor responses (change in discharge activity in response to passive or active movement or tactile stimulation).
Data Collection and analysis. We tested the accuracy of Cicerone by having one experimenter map the basal ganglia using the software while another experimenter simultaneously used the conventional method of plotting recording sites on graph paper.
We evaluated how well the recording data matched Cicerone’s custom atlases (warped
3D atlas and linearly scaled 3D atlas) compared to two conventional atlases (standard 3D atlas and print atlas). For monkey 04-m-005 who had mapping data recorded in the coronal plane, we used the Paxinos et al. (1999) print atlas. For monkey 04-m-001 we used the sagittal Ilinsky et al. (2002) print atlas. The standard 3D atlas was built from the
Martin and Bowden (2000) digital atlas as described above. Since their original 2D atlas was made from a longtailed macaque, we scaled it to better approximate rhesus brain using the average scaling factors from Martin and Bowden (2000) (1.08 in AP, 1.06 in
53 DV and 1.0 in ML direction). As a result the standard 3D atlas was adapted to the rhesus
species, but not customized to either animal.
For each atlas we calculated the number of recording sites that were within their
proper nuclei, and for sites that were outside their nucleus we calculated the number of
recording sites that were within 1 mm from the border of their proper nucleus.
Recordings had to be aligned with each atlas, so all the atlases were shifted to properly match the GP recording data with the GP atlas structure in monkey 04-m-005, and the
STN recording data with the STN atlas structure in monkey 04-m-001. The GP and STN recordings were selected in the respective animals because these were the target nuclei for DBS electrode implantation. Initially, the atlas AC was aligned with the MRI-defined
AC. However, to achieve the proper fit with the neurophysiologic data (GP or STN), the
3D atlases had to be shifted within the skull by varying amounts reported in the Results section (see also Discussion). Track depths were also adjusted in both monkeys. In monkey 04-m-001, 6 tracks (all recorded with the same electrode) were shifted by 2 mm, and in monkey 04-m-005, multiple shifts were performed averaging 0.6 mm. For print atlas accuracy analysis we selected the nearest available atlas slices that best fit the data
(allowing for track depth adjustments), and overlaid the atlas transparencies with MER data plotted on the graph paper. We tested the differences in the total numbers of MER sites inside the nuclei, < 1 mm outside and > 1 mm outside between the four atlases using chi-square test with Bonferroni correction for multiple comparisons. A significant difference was accepted when p-value was < 0.01.
54 2.3. Results
We used the Cicerone software system to plan the implantation of recording and
stimulation chambers in two rhesus monkeys and to visualize MER data with respect to
the surrounding basal ganglia neuroanatomy. The accuracy of Cicerone’s subject-specific
atlases (warped and linearly-scaled) and the standard atlases (3D and print) was evaluated
by comparing how well neurophysiologically defined MER locations matched the atlas-
defined nuclei. Data entry using Cicerone was performed at equal speed as taking paper notes and it did not slow down the mapping process. An additional advantage of Cicerone was that the data did not have to be recopied for entry into a computer database.
In monkey 04-m-005, 14 electrode penetrations were made (data from two tracks were discarded because they were deemed unreliable) in 5 coronal planes and a total of
1271 locations with extracellular unit activity were recorded. Of those, 579 were classified as belonging to striatum (63%), GPe (10%), GPi (10%), or optic tract (16%)
(Fig. 2-2 A-D). Figure 2-3A shows the percentage of classified recording sites and their position with respect to the atlas-defined nuclei. The total percentage of MER sites within their proper nuclei was 73% for Cicerone’s warped 3D atlas, 72% for Cicerone’s linearly- scaled 3D atlas, 63% for the standard 3D atlas and 61% for the print atlas. The warped atlas had to be moved an average of 0.5 mm to align the GP MER data with the atlas- defined structures (0.15 mm medial, 1.15 mm anterior and 0.1 mm ventral). For the linearly-scaled atlas the average was 1.2 mm (1.4 mm lateral, 0.85 mm anterior and 1.3 mm ventral) and 1.6 mm for the standard 3D atlas (2.9 mm lateral, 0.45 mm posterior and
1.45 mm ventral). The corresponding translations were difficult to estimate for the print
55
Figure 2-2. Microelectrode recording data and Cicerone atlas from monkeys 04-m-004 (A–D) and 04-m-001 (E–H). (A) MER sites defined as GPe or GPi are displayed as green or blue spheres within the GPe and GPi structures of Cicerone’s warped 3D atlas. (B) The chamber and MER data from four recording tracks are shown within a coronal slice of Cicerone’s warped 3D atlas. (C) Zoomed in view of (B). (D) Same four tracks as in (B) and (C) displayed within the standard 3D atlas. MER sites from putamen (red), caudate (yellow), GPe (green), GPi (blue), STN (maroon), optic tract (pink), electrophysiologically quiet sites without a visual stimulation response (black), border cells (brown), and unclassified (white) are displayed with markers. The marker shape is determined by the presence (cylinder, cube or cone) or absence (sphere) of a sensorimotor response. Caudate and putamen sites are distinguished solely by their inferred anatomical location. (E) STN and GPi MER sites are shown within transparent nuclei of Cicerone’s warped 3D atlas. (F) Five MER tracks in an oblique slice (15◦ from sagittal plane) from Cicerone’s warped 3D atlas. (G) Zoomed in view of (F). (H) Same five tracks as in (F) and (G) displayed within the standard 3D atlas. Each leg of 3D scale bars represents 3mm (D: dorsal; P: posterior; M: medial).
atlas since the chambers were rotated and translated in a 3D space that cannot be captured with a 2D print atlas.
In monkey 04-m-001, there were 16 electrode tracks in 7 oblique planes and a total of 1480 recording sites, of which 417 were classified as striatum (54%), GPe (18%),
GPi (7%), STN (17%), or optic tract (4%) (Fig. 2-2 E-H). The total percentage of MER
56 sites within their proper nuclei was 79%, 66%, 42% and 48% for Cicerone’s warped 3D atlas, Cicerone’s linearly-scaled 3D atlas, the standard 3D atlas and the print atlas, respectively (Fig. 2-3B). The warped atlas had to be moved an average of 0.8 mm to align the STN MER data with the atlas-defined structure (0.75 mm lateral, 1.6 mm
Figure 2-3. Cicerone atlas accuracy for monkey 04-m-005 (A) and 04-m-001 (B). Percentage of MER sites that were located within their atlas-defined nucleus, less than 1mm from the nucleus border or more than 1mm from the nucleus border for Cicerone’s warped custom 3D atlas (W), Cicerone’s linearly scaled custom 3D atlas (L), Cicerone’s standard 3D atlas (S) and 2D print atlas (P). Connecting lines indicate significant difference between the atlases. Total numbers in the three categories (inside nucleus, <1mmoutside, >1mmoutside) for each atlas were compared to the warped atlas totals using the chi-square test with Bonferroni correction (p < 0.005). Since striatum sites made up more than half of all MER sites and were therefore overrepresented in the totals, calculations for significance were repeated without the striatum sites (indicated with asterisks). Statistical analysis of the pooled data (totals for both monkeys added) yielded the same significant differences between the atlases as shown in (B).
57 anterior and 0.15 mm ventral). For the linearly-scaled atlas the average was 1.0 mm (1.1 mm lateral, 1.25 mm anterior and 0.7 mm ventral) and 1.2 mm for the standard 3D atlas
(1.95 mm lateral, 0.65 mm anterior and 1.0 mm ventral).
The DBS electrodes were implanted in the motor regions of the posterior GP and lateral STN, respectively, as defined by the neurophysiologic data. Figure 2-1C shows the theoretical DBS VTA for monkey 04-m-001 which was used to assist in deciding the optimal electrode position. A post-implantation CT confirmed the electrode locations and preliminary results show therapeutic effects with stimulation suggesting that the electrodes were placed within the desired nuclei.
2.4. Discussion
This study evaluated the utility of a 3D visualization and database software system for stereotactic neurosurgical planning and neurophysiological microelectrode recordings in non-human primate research. The purpose of the Cicerone system is to aid with head chamber positioning and provide visual feedback during the microelectrode recording procedures. Cicerone also provides guidance in DBS electrode implantation by estimating the volume of tissue activated for a given DBS electrode position and stimulation parameter setting. Further, the database components of the system simplify the neurophysiological record keeping process and allow for advanced data analysis and visualization. We demonstrated Cicerone’s capabilities by analyzing the match between subject-specific 3D atlases and MER data in two parkinsonian rhesus monkeys undergoing GP and STN DBS implantation. 3D atlases customized to the MRI of the
58 animal by either non-linear warping or linear scaling provide a better match with MER data than either a standard 3D atlas or a print atlas. The limitations of various techniques integrated into the software are discussed below.
The standard 3D atlas used in this study, created from a longtailed macaque (M. fascicularis) digital template atlas (Martin and Bowden, 2000), was adapted for use in rhesus macaque (M. mulatta). Analysis by Bowden and Dubach (Ch 5. in Martin and
Bowden, 2000) using published atlases of various primate species concluded that with appropriate linear scaling the template atlas was suitable for stereotactic targeting in several other species of macaque (M. mulatta, M. fuscata) and baboon (P. papio, P.
cynocephalus, P. anubis). Their analysis consisted of first calculating the mean linear
ratios for each species by comparing positions of 13 brain landmarks in the template atlas
and in the atlases of given species. The mean linear ratios for each species were then used
to scale the template atlas. The authors calculated the discrepancy between the landmark
positions in the scaled template atlas and the actual atlas for that species. Absolute
discrepancy of 0.5 mm or less was considered acceptable. It should be noted that this
comparative analysis was not performed for cortex and the lower brain stem so the scaled template atlas would not necessarily be suitable for use in these brain areas.
While the appropriately scaled M. fascicularis template atlas can be used for similar species with reasonable accuracy, inter-individual variability is still a limiting factor for atlas based stereotactic targeting (Percheron and Lacourly, 1973; Francois et al., 1996;
Deogaonkar et al., 2005; Martin and Bowden, 2000). Our results show that the standard
3D atlas (scaled using mean linear ratios for M. mulatta) provided limited agreement between the atlas and MER data. Customization of the 3D brain atlas to the MRI of the
59 specific animal significantly improved the accuracy of the anatomical match of the model to the MER (Fig. 2-3). Interestingly, simple linear scaling of the 3D brain atlas to match the MRI generated an effectively equal match for our target nuclei, compared to the more difficult and time consuming non-linear warping.
Unlike human stereotactic neurosurgery, image guidance is less commonly employed in non-human primate procedures. Our experience suggests that the integration of pre-operative imaging and 3D visualization software can provide invaluable assistance in surgical planning and data collection in non-human primates. Our goal was to position two metal chambers (19 and 15 mm diameter) on small skulls, approximately 70 mm long and 50 mm wide, such that the chambers did not interfere with each other and that microelectrode recording from selected subcortical structures could occur without interference from the implanted DBS lead. After importing a CT scan and visualizing the skull, Cicerone’s planning tools made it easy to move and rotate chambers in 3D virtual space and find the optimal positions prior to surgery. Cicerone can also be used with just an MRI where bony landmarks are defined using an MRI-compatible stereotactic frame.
However, a CT provides much better skull visualization and it does not introduce artifacts from head chambers, skull screws and head posts implanted on the skull. In its simplest form Cicerone can also be used without any imaging data.
Additional problems that confront every neuronavigation system are imaging and registration accuracy. Imaging issues such as voxel size and image distortion have been extensively analyzed (Sumanaweera et al., 1994; Hardy and Barnett, 1998; Poggi et al.,
2003). The animal procedures provide an additional challenge because two imaging modalities must be co-registered, an MRI used to create a subject-specific atlas and a CT
60 used to visualize external bony landmarks that define the stereotactic frame space. This provides yet another potential source of error which can be partially compensated for by adjusting the CT/MRI fusion and/or position of the 3D atlas within the skull. Despite the list of issues that limit stereotactic accuracy, the use of image-guided tools, such as
Cicerone, provide substantial improvements over currently used techniques in non-human primate research.
MER data acquisition using Cicerone was simple, as fast as using paper records, and it allowed for immediate visualization of recording locations in the context of the 3D neuroanatomy. Two potential problems that can affect the accuracy of MER mapping are electrode deflection and brain shift (Finnis et al., 2003). Cicerone assumes that the electrode is following a perfectly straight virtual trajectory through the brain, when in fact deviations in the trajectory may occur. The degree of error is hard to estimate. The CT scans following the DBS electrode implantation in one of our monkeys showed a slight curving of the electrode trajectory throughout its length which resulted in a ~0.75 mm deflection at the tip. The DBS electrode is flexible and the curvature may have resulted from a number of factors. Another source of error that should be recognized is the microdrive guide tube. The relationship between the guide tube and the microdrive electrode attachment may not be perfectly aligned and result in an improper electrode penetration angle. In fact we noticed this problem with our equipment, prompting us to custom-design a metal stage for the microdrive for future studies.
Deformation of the brain during open-skull neurosurgical procedures can be a significant source of error for neuronavigation systems (Hastreiter et al., 2004). Brain shift in human surgeries increases with time, so microrecording trajectories obtained
61 early in the procedure are more likely to align with atlases customized to the preoperative
MRI/CT scans. This problem is likely even more pronounced in animal procedures where
permanent craniotomies are used and microelectrode mapping occurs over a period of
several days. It is unfortunately not useful to obtain an MRI following the craniotomy
because of the large artifact caused by the metal chambers and skull screws, so we must
rely on the image taken when the animal’s skull was intact. In addition, during the MRI
scan the animal is in the supine position, but during MER data collection the animal is
sitting upright in the primate chair which might cause a slight anterior/posterior shift in
the predicted electrode position (see below) (Rohlfing et al., 2003).
We therefore found it necessary to provide Cicerone users with the option of
repositioning the 3D atlas to better match the MER data. For the purpose of evaluating
the system accuracy we moved the 3D nuclei using only the mapping data that defined
our target structures, the GP in monkey 04-m-005 or STN in monkey 04-m-001. In
practice, we reposition the 3D atlas after the first few recording tracks; thereby providing
a better estimate for the subsequent recordings. In both animals we found that Cicerone’s
warped 3D atlas had to be shifted anteriorly by 1-1.5 mm and ~0.5 mm in the other two
directions. These shifts attempt to provide correction for all potential errors (electrode
deflection, brain shift, microdrive imprecision, size and shape of 3D nuclei, MRI/CT co-
registration), but it is difficult to estimate how much each problem contributes to the
overall inaccuracy. Nonetheless, the overall ~1 mm accuracy is within the range of
clinical stereotactic systems (Maciunas et al., 1994). Another limitation that should be
recognized is that several microelectrodes are typically used during the mapping
procedure. It is difficult to zero each electrode to precisely the same location, which can
62 cause the depths of the individual mapping tracks to be slightly offset from each other.
Cicerone enables the user to move each mapping track independently to align them
properly, usually using a well defined anatomical region present in the track.
A novel feature of Cicerone is the ability to visualize the estimated volume of
tissue activated (VTA) by the DBS electrode. This offers a great potential benefit to
researchers who could anticipate stimulation effects before the actual electrode
implantation, and plan the electrode trajectory to achieve the desired interaction between the VTA, MER data and 3D neuroanatomy. In this study we concentrated on stimulation
models customized to the scaled clinical DBS system used in the Vitek laboratory
(Hashimoto et al., 2003; Elder et al., 2005). However, future versions of Cicerone can be
easily adapted to any given electrode design and stimulation paradigm. The VTA
predictions are based on theoretical models of the neural response to extracellular electric
fields (Butson and McIntyre, 2005; 2006; Butson et al., 2006a,b; Miocinovic et al., 2006).
Our VTA predictions incorporate the latest advances in neural stimulation modeling such
as explicit characterization of capacitance and impedance of the electrode-tissue interface
and detailed multi-compartment biophysical models of myelinated axons. Assumptions
used to make the VTA predictions in Cicerone are that all the axons were of the same
diameter, axonal trajectories and orientations were uniform, and bulk tissue conductivity
was homogeneous. While all these factors influence the spread of stimulation, certain
trade offs were necessary to reduce the computational complexity and limit the memory
requirements for the Cicerone software system.
Experimental validation of the VTA models is a difficult task. We are actively pursuing
research studies that link our model predictions with electrophysiological recordings in
63 humans and non-human primates (Miocinovic et al., 2006; Butson et al., 2006b). The results of these studies show that our models can accurately predict stimulation spread into the corticospinal tract during STN DBS. In turn, it is currently possible to use neurostimulation models to make quantitative, experimentally relevant, predictions. And, while the current VTA prediction functions in Cicerone are relatively simple, we believe that the synergistic evolution of our modeling technology and experimental analysis will allow for continuous improvement in their accuracy and validity.
Our results show that Cicerone has the potential to significantly increase the accuracy and precision of stereotactic implantation of neuromodulatory devices. But it should be noted that the utility of Cicerone will depend on the users’ skill in customizing the 3D atlas to animals’ MRI/CT, experience in characterizing neuronal firing properties from the MER data, and precision in making atlas/track shift adjustments. Although the
Cicerone system used in this study requires additional software components (Analyze,
Edgewarp) to be fully functional, we are currently in the process of further refinements that will result in a stand-alone software system. Our intention is to provide Cicerone to the non-human primate scientific research community free of charge.
64 Chapter 3: Spatial and temporal characteristics of the voltage field generated by deep brain stimulation electrodes
3.1 Introduction
One of the greatest obstacles in studies aimed at elucidating mechanisms of DBS has been the inability to accurately predict the spread of stimulation and define the specific brain areas affected by the applied electric field. As a result neural structures responsible for therapeutic effects of DBS have not been clearly identified, and this limitation severely hampers efforts to explain its mechanisms of action. A number of studies have attempted to qualitatively describe therapeutic target area by determining the anatomical location of the active electrode contact (Starr et al., 2002; Lanotte et al., 2002;
Saint-Cyr et al., 2002; Voges et al., 2002; Hamel et al., 2003; Yelnik et al., 2003; Herzog et al., 2004; Zonenshayn et al., 2004; Nowinski et al., 2005; Yokoyama et al., 2006;
Godinho et al., 2006). However, the volume of tissue affected by stimulation will also depend on stimulation parameters (amplitude, pulse width, frequency, polarity), electrode characteristics, and electrical properties of the surrounding tissue. Without considering these additional factors, it is impossible to determine if stimulation effects will be contained to the anatomical region of the active contact or if they will extend to the surrounding areas.
The response of neural cells to DBS occurs as a result of voltage field induced in the tissue by the electrical stimulation. Local voltage field gradients cause ionic current to flow through the cell membrane. The direction of current flow determines whether the membrane becomes depolarized or hyperpolarized. To predict how far and in what way
65 the neurons are affected by stimulation, it is first necessary to estimate voltage generated in the tissue by the electrode. This problem can be addressed by constructing volume conductor models, in which tissue is assigned electrical conductivity value(s) and the
Poisson equation is used to solve for electrical potentials in the space. The calculated voltages can then be applied to model neurons distributed in the space around the electrode to predict the neural response to stimulation (McNeal, 1976). This approach has been utilized in many neural engineering research areas, such as stimulation of the spinal cord (Coburn and Sin, 1985; Struijk et al., 1991; Holsheimer, 1998; Rattay et al., 2000), cortex (Grandori and Rossini, 1988; Manola et al., 1995), heart (Latimer and Roth, 1998), peripheral nerves (Sweeney et al., 1990; Hoekema et al., 1998 Choi et al., 2001), optic nerve (Oozeer et al., 2005), auditory nerve (Ruddy and Loeb, 1995; Hanekom et al.,
2005) and limbs (Reichel et al., 2002).
Volume conductor models have also been applied to DBS research (McIntyre et al., 2004a, 2004b; Butson and McIntyre, 2005; Hemm et al., 2005; Astrom et al., 2006;
Gimsa et al., 2006). Recently, we have used DBS volume conductor models to construct detailed field-neuron models in an attempt to quantify spread of stimulation and identify therapeutic target areas (Miocinovic et al., 2006; Butson et al., 2007). The volume conductor models, however, have not been validated in vivo, and there exist no direct behavioral or functional correlates that could be used to adequately assess the accuracy of the DBS electric field models. As a result this study was undertaken with the goal of characterizing spatial and temporal characteristics of the voltage field generated in tissue by DBS electrode. The experiments were performed by recording voltages around an active DBS electrode in a saline bath or implanted in the thalamus of a monkey. These
66 experimental data were then compared to volume conductor models parameterized to match the different experimental environments.
3.2 Materials and methods
Voltage induced in tissue or saline by a DBS electrode was recorded with a microelectrode positioned at various distances from the active DBS electrode. By varying vertical and horizontal distance, multiple recordings were obtained and voltage field maps were constructed. The temporal characteristics of induced voltages were observed for both constant voltage (voltage-controlled) and constant current (current-controlled) stimulation. Volume conductor models with homogeneous isotropic bulk tissue mediums were constructed to represent the voltage fields and stimulation waveforms recorded in vivo and in vitro.
3.2.1 Stimulation and recording protocols
DBS electrodes used in this study were scaled-down versions of the clinical DBS electrodes, suitable for implantations into the brains of non-human primates (Hashimoto et al., 2003; Elder et al., 2005). The electrode consisted of a 45 mm long shaft with four cylindrical contacts wrapped around the distal end of the lead. The contacts were 0.75 mm in diameter, 0.5 mm in length and separated by 0.5 mm of insulation spacing. The electrode contacts were made from platinum/iridium alloy, and the shaft was polyurethane. The electrodes were manufactured by the Advanced Bionics Corporation
(Valencia, CA).
67 Both constant voltage and constant current pulses were used for stimulation through the DBS electrodes. For constant voltage stimulation we used a biphasic pulse generator (BPG-1; Bak Electronics, Mount Airy, MD) and a biphasic constant voltage stimulus isolator (BSI-1; Bak Electronics). For constant current stimulation we used a pulse generator (S88; Grass Instruments, Quincy, MA) and two photoelectric constant current stimulus isolation units to achieve biphasic waveform (PSIU6; Grass
Instruments). Stimulation parameters were 0.3 V or 30 µA amplitude, 500 µs cathodic
pulse, 500 µs interpulse delay, and 500 µs anodic pulse (Fig. 3-6). The biphasic pulses
were repeated at 20 Hz. Stimulation was monopolar, where the electrode contact served
as a cathode while anode was a large distance away. The stimulus amplitude was chosen
so that it was low enough to be subthreshold for any behavioral effects in the animal. The
pulse width and interpulse delay were chosen so that they are long enough to observe
temporal characteristics of the recorded waveform. The frequency was chosen to be low
enough so that the recorded voltage could return to baseline between pulses and so that
any activation of nearby neurons did not affect recorded amplitude of the subsequent
pulse (20 Hz = 50 ms between biphasic pulses). Stimulation was applied for 5 seconds at
each DBS contact for each microelectrode recording position. The active contact was
selected with a switch so responses at all four contacts were recorded in 23 seconds (3
seconds allowed for manual switching between the contacts).
Voltages induced in the tissue or saline by a DBS electrode were measured with a
microelectrode to achieve fine spatial resolution. Specific details regarding in vitro and in
vivo measurements are described in subsequent sections. In both cases epoxylite-coated
tungsten microelectrodes with tip lengths of approximately 50 µm (FHC, Bowdoinham,
68 ME) were positioned at different vertical and horizontal distances from the DBS electrode using a microdrive (MO-95-lp, Narishige Scientific Instruments, Tokyo, Japan).
Vertical spatial resolution was 0.25-0.5 mm while horizontal resolution was 0.5-1 mm.
The recorded signal was amplified (50x) and bandpass filtered (0.1 Hz – 20 kHz) using a differential amplifier with a headstage (A-M systems model 3000, Sequim, WA). The signal was then digitally sampled at 100 kHz and stored for later offline analysis (Power
1401 and Spike2 software, Cambridge Electronic Design, Cambridge, UK). Voltage field spatial maps were constructed by plotting the average peak cathodic voltage amplitude measured at multiple locations in the same (or parallel) plane where the DBS electrode was located. The maps were interpolated in Matlab (Mathworks, Natick, MA).
Monitoring electrode impedance is important both for understanding the stimulation effects and recorded responses. DBS electrode impedance influences voltage amplitude induced in the medium and microelectrode impedance affects amplitude of the recorded signal. DBS impedance recordings were performed using 1KHz sinusoid impedance tester (IMP-1, Bak Electronics, Mount Airy, MD), and these measurements were used to classify the various microelectrode recording sessions (see Results).
Microelectrodes were regularly tested to assure their impedances remained constant during a series of experimental voltage recordings (the same microelectrode was usually used for several days). The goal was to minimize potential differences in recorded voltage fields that would result from microelectrodes with slightly different impedances and different positions within the microdrive. Increases in microelectrode impedance would usually signal occlusion at the tip, such as a piece of tissue, which could usually be carefully removed and the electrode could be reused. Unexpected decrease in impedance
69 would indicate that the electrode was probably damaged, and it would be replaced. Initial
microelectrode impedances were between 0.6-1 MΩ, and daily variations within ±0.2MΩ
were tolerated.
3.2.2. In vitro experiments
In vitro experiments were performed by suspending a DBS electrode in saline.
Constant voltage or constant current stimulus pulses were applied through the DBS
electrode and the induced voltage field was recorded with a microelectrode as described
in the previous section.
The cylindrical glass jar was 7 cm in diameter, 7 cm in height and filled with
0.9% NaCl (conductivity 1.5 S/m). The DBS electrode was affixed to the jar lid so that
the electrode was vertical and its contacts were approximately in the middle of the jar. A
stainless steel wire was wound around the inner wall of the container and served as the return electrode. A silver/silver chloride wire was positioned several centimeters from the recording microelectrode and it served as the reference electrode (Fig. 3-1A).
Photographs were taken during the recordings to verify microelectrode distance from the
DBS electrode.
3.2.3 In vivo experiments
A DBS electrode was chronically implanted into ventral thalamus of a rhesus monkey (Elder et al., 2005). Constant voltage or current stimulus pulses were applied through the DBS electrode and the induced voltage field was recorded with a microelectrode as described in the previous section.
70 One female rhesus monkey (M. mulatta; ID number 05-m-003) weighing 5.0 kg was used in the study. All surgical and recording protocols were approved by the
Cleveland Clinic Institutional Animal Care and Use Committee and complied with
United States Public Health Service policy on the humane care and use of laboratory animals.
Imaging
Magnetic resonance imaging (MRI) was used to obtain anatomical images for surgical planning. Images were acquired on Siemens MAGNETOM Trio 3 T scanner
(Siemens Medical Systems, Erlangen, Germany), while the animal was under propofol anesthesia. T2-weighted images consisted of twenty 2 mm thick coronal slices and twenty 2 mm thick sagittal slices (256×256 matrix, 120x120mm field of view (FOV),
0.47x0.47mm in-plane resolution).
Computed tomography (CT) scans, performed with a Siemens SOMATOM
Sensation, were used for surgical planning and subsequent localization of the implanted
DBS electrode (Fig. 3-2 D E). CT images were acquired in the axial plane in 1mm increments (135 slices at 512×512 pixels; 0.24mm×0.24mm in plane resolution). MRI and CT images were coregistered in Analyze 7.0 (AnalyzeDirect, Lenexa, KS).
Surgical procedure
A recording chambers was implanted in an aseptic surgical procedure under isoflurane anesthesia with the head held in a primate stereotactic frame (Kopf frame model 1430). Prior to surgery MRI and CT images were imported into our neurosurgical
71 navigation software Cicerone (Miocinovic et al., 2007) and stereotactic chamber coordinates calculated (Fig. 3-2A). A craniotomy was performed and one metal chamber
(16 mm inner diameter) was anchored over the left hemisphere. The chamber targeted the thalamus, and was placed in the sagittal plane at 5° from the midline (anterior-to- posterior), 6.6 mm anterior and 5.4 mm lateral (in frame coordinates). A post-operative
CT verified that the chamber was implanted as planned.
Neurophysiological mapping and DBS electrode implantation
The DBS electrode was implanted in a separate procedure after several days of neurophysiological mapping of the target brain region (Elder et al., 2005; Miocinovic et al., 2007). Extracellular neuronal activity was recorded though the chambers targeting ventral thalamus to find the optimal position for the electrode implantation. During microelectrode mapping and DBS implantation the animal was sitting in a primate chair with the head stabilized with a metal head post. During DBS implantation the animal was sedated with ketamine (10 mg/kg) to prevent any movement. The recording equipment was the same as described above for voltage field measurements except to detect neuronal activity the recorded signal was amplified 1000x, bandpass filtered at 500–3000
Hz (Krohn-Hite model 3380, Brockton, MA), and digitally sampled at 25 kHz.
Voltage field data collection
During microelectrode voltage field recordings the animal was lightly sedated with acepromazine (1 mg/kg) and sitting in a primate chair with the head restrained. As a result, the animal was awake but drowsy and the amount of voluntary movement was
72 significantly reduced thus minimizing recording noise and pressure exerted on the head implant. The recording sessions lasted up to 4 hours. The animal was also receiving 4 mg of prednisolone daily to treat an endogenous intestinal disorder.
Microelectrode voltage field recordings were performed through the same chamber where the DBS electrode was implanted. The microelectrode was positioned parallel to the DBS electrode, either in the same sagittal plane or sagittal plane 1 mm away. The microelectrode guide tube inserted a few millimeters below the dura was used as a reference electrode. A chamber on the opposite hemisphere was used for stimulation current return.
X-ray images were acquired for each recording track to verify the distance from the microelectrode to the DBS electrode (Fig. 3-2 C D). A portable veterinary unit (PXP-
40HF; Poskom, Korea) was used with 10x12 inch film and an intensifying grid cassette.
The X-ray source was set to 70 kVp and 2 mA and positioned 106 cm from the film cassette. A custom-made alignment tool was used to position the animal so that the images were taken in a sagittal plane (the same plane in which the DBS electrode and microelectrode were inserted). During a single recording session the X-ray source, film cassette, and animal were kept in the same position, so that only the microelectrode position changed. However, small variations in day to day positioning were unavoidable.
Films were digitized and distance of the microelectrode to the active contact was measured in Adobe Photoshop v7.0.
73 3.2.4 Modeling studies
Spatial voltage field model (static solution)
An axisymmetric finite element model (FEM) of the DBS electrode was created in FEMLAB v3.1 (COMSOL, Burlington, MA) to calculate potentials generated in the medium by the electrode. The volume conductor was 5x5 cm in size and grounded at the boundaries. The electrode shaft was modeled as an insulator (1e-6 S/m) and each electrode contact as a conductor (1e6 S/m). An encapsulation sheath surrounded the electrode shaft in the models of in vivo DBS voltage fields (Butson et al. 2006; Grill and
Mortimer 1994).The bulk conductivity of the tissue medium and encapsulation sheath were obtained from parameter estimation calculations. In the models of in vitro DBS voltage fields, saline conductivity was set to 1.5 S/m and no encapsulation sheath was present. The electrode-tissue interface was modeled as a thin film approximation in
FEMLAB. Thickness and conductivity of the interface layer were also obtained from
parameter estimation. The model had variable resolution mesh with a total of 32,640
mesh elements that increased in size with increasing distance from the electrode. Voltage
values within the volume were determined from the Poisson equation, which was solved
using direct matrix inversion (UMFPACK solver). Model impedance was calculated by
dividing the applied voltage by the applied current, where the applied current was
calculated by integrating the current density over the electrode contact surface area.
74 Parameter estimation
A least-squares parameter estimation method was implemented in Matlab to
optimize model parameters to fit the in vivo and in vitro voltage field data. The initial
parameter values were 0.2 S/m for tissue conductivity (Ranck 1963), 0.1 S/m for encapsulation conductivity and 0.25 mm for encapsulation thickness (Grill and Mortimer
1994), 0.1 S/m for interface conductivity and 0.1 mm for interface thickness.
Temporal voltage field model (time-dependent solution)
The stimulus waveforms produced by the stimulation pulse generators were square biphasic pulses. However, the actual stimulus waveform delivered to the medium is modified by the capacitance at the electrode-tissue interface and/or capacitance of the bulk tissue medium (Butson and McIntyre, 2005). For voltage-controlled stimulation, electrode capacitance plays a dominant role in this phenomenon, whereas for current- controlled stimulation tissue capacitance is more important. A previously developed
Fourier FEM was used to account for the effects of electrode and tissue capacitance
(Butson and McIntyre, 2005). The general steps of this method are briefly described here.
First, the stimulation pulse generator waveform was constructed in the time domain. It was converted to the frequency domain using a discrete Fourier transform (DFT) in
Matlab. The Fourier FEM was solved at each component frequency of the DFT (1,024 frequencies between 0 and 50 kHz). The result at each frequency was scaled and phase shifted using the DFT magnitudes and phases. Finally, an inverse Fourier transform was performed to obtain the stimulus waveform in the time domain. For voltage-controlled stimulation electrode capacitance was set to 0.6 µF, a value estimated from in vitro
75 recording data. The model of in vitro voltage distribution used a saline dielectric constant
of 50. The model of in vivo voltage distribution used a dielectric constant of 3*106 which
is in the upper range of values found in the literature (1*104 to 1*106; Butson and
McIntyre, 2005); however, this value produced waveforms most similar to those recorded
in vivo.
Figure 3-1. In vitro voltage field recordings. A. In vitro recordings were performed with a four-electrode setup. A voltage or current stimulus was applied across the DBS electrode and a return electrode. The response voltage was measured between the microelectrode and reference electrode. B. Peak cathodic voltage was recorded at different depths 1 mm away from DBS electrode. The four curves correspond to independent activation of the four different contacts. C. Voltage field map obtained by interpolating recordings at 28 depths and 6 distances (168 total locations) from the DBS electrode (individual recording locations are marked with small black dots).. D. Model solution optimized to fit the in vitro recording data (electrode-saline interface thickness 0.1mm, interface conductivity 0.37 S/m). 76 3.3 Results
3.3.1 In vitro voltage recordings
An in vitro recording setup (Fig. 3-1A) was used to measure voltages generated in
saline by a DBS electrode. Figure 3-1B shows peak cathodic voltages recorded at
different depths, 1 mm away from DBS electrode. Recordings were done at 0.25 mm
increments. Four curves correspond to the independent activation of the four different
contacts. Recorded voltage increased as the microelectrode tip approached the active
contact and decreased as it moved away. The spatial map of peak voltage distribution
around the DBS electrode is shown in Fig. 3-1C.
3.3.2 In vitro voltage distribution model
Figure 3-1D shows a spatial map of voltages calculated from the FEM at the same
locations that the in vitro data was collected. The model parameters were optimized to fit
the in vitro data. The conductivity of saline was set at 1.5 S/m, and the two remaining
parameters, electrode-saline interface thickness and conductivity, were estimated. It was
found that only the ratio of the two parameters needed to be estimated to find a unique
solution. Therefore, the interface thickness was set to 0.1 mm and the optimal interface
conductivity was found to be 0.371 S/m (0.361-0.382 confidence interval (CI)). The
resulting model impedance was 1.03 kΩ, compared to 0.75 kΩ recorded in-vitro. Since
impedance decreases with frequency and experimental impedance was recorded at 1kHz it is expected that the experimentally measured impedance was lower than the impedance calculated for the static model (0Hz).
77
Figure 3-2. Surgical planning and electrode implantation. A. Cicerone software was used to plan a recording chamber location that targeted the thalamus. B. X-ray image taken during a recording session. C. Zoomed in view of the X-ray image of the DBS electrode and recording microelectrode. The microelectrode was 2.1 mm from the DBS electrode. D. the post-DBS-implantation CT was coregistered with pre-DBS-implantation MRI to define the DBS electrode location in the brain. The white structure is electrode contour extracted from the CT data. E. Cicerone was used to visualize electrode with respect to a brain atlas warped to match the neuroanatomy of the animal.
3.3.4 In vivo voltage recordings
The DBS electrode was chronically implanted in the ventral thalamus of a monkey (Fig. 3-2). Figure 3-3 shows voltages recorded in the same microelectrode locations over several days post-DBS-implantation. On the first day after implantation, recorded voltages were larger than on the subsequent days, and electrode impedance recorded on the first day was lower than in the following two weeks (3 kΩ vs. 5.6±1 kΩ).
Reported impedances were measured after stimulation through the DBS electrode.
Stimulation reduced electrode impedance and this effect was most significant during the first 5-10 minutes of the stimulation. Since electrode impedance varied over time, the
78 results are divided into three groups: low (Fig. 3-4A), medium (Fig. 3-4C) and high (Fig.
3-4E) impedance. The low impedance data consists of 3 tracks recorded in one day when
the electrode impedance was 2.4 kΩ. The medium impedance voltage map includes 7
tracks recorded on two separate days when impedance was measured at 4.3 kOhm and
4.7 kΩ. The high impedance data contains 4 tracks measured in one day when impedance
was 8 kΩ. Figure 3-5 summarizes all the recorded tracks and it can be observed that the
voltage amplitude depends both on distance from the DBS electrode and impedance of
the DBS electrode.
Figure 3-3. Repeated in vivo DBS voltage measurements. Voltage field recordings were repeated over several days at the same microelectrode locations (2 mm from DBS electrode).
Figure 3-4. In vivo voltage field recordings. On the left side are voltage maps obtained from in vivo recordings on 4 days (middle row combines 2 days’ data). Black dots indicate recording locations – microelectrode penetrations were made in the same sagittal plane where the DBS electrode was implanted. On the right side are model solutions optimized to fit the in vivo recording data. DBS electrode impedance varied daily so recording data was divided according to particular day’s impedance value: top row is low impedance (2.4 kΩ in vivo, 9.4 kΩ model), middle row is medium impedance (4.3 and 4.7 kΩ in vivo, 15.7 kΩ model) and bottom row is high impedance (8 kΩ in vivo, 20.5 kΩ model). Color bar uses logarithmic scale.
79
80 3.3.5 In vivo voltage distribution model
DBS FEMs were constructed and optimized to match the low, medium, and high
impedance in vivo data sets. Each model consisted of a DBS electrode, surrounded by a
tissue encapsulation layer and a homogeneous isotropic bulk tissue medium. Tissue
conductivity was set to 0.2 S/m and encapsulation thickness to 0.25 mm for all three
models. The calculated optimal encapsulation conductivity for the low, medium, and high
data sets were 0.0334 S/m (0.0308-0.0361 CI; Fig. 3-4 top right), 0.0186 S/m (0.0181-
0.0191 CI; Fig. 3-4 middle right), and 0.0139 S/m (0.0136-0.0142 CI; Fig. 3-4 bottom
right), respectively. The model impedances were 9.4, 15.7 and 20.5 kΩ, respectively.
Since this was a static model, the impedances are higher than reported for the in vivo data
where measurements were made at 1kHz.
Figure 3-5. Effect of DBS electrode impedance on recorded voltage. Each point represents one in vivo recording track (14 tracks total) and the largest voltage recorded in a track is shown. Recording tracks are separated according to the impedance measured post-stimulation on the given recording day.
81 3.3.6 Stimulation waveform characteristics
Spatial voltage maps display only the peak voltage recorded during the time the
stimulus pulse was applied. The voltage response in fact varied over time as
demonstrated in Fig. 3-6. In the case of voltage-controlled stimulation the peak voltage
occurred at the beginning of the pulse and the response decayed over time (that is less
current was injected over time). For current-controlled stimulation, the recorded
waveform more closely followed applied pulse, with the peak at the end of the pulse.
For both types of stimulation, the waveform generated by the Fourier FEM model
closely matched the experimentally recorded waveform. The in vitro voltage distribution
model used the following parameter values: saline conductivity 1.5 S/m, saline dielectric
constant 50, and electrode capacitance 0.58 µF (estimated from in vitro time constant).
The in vivo voltage distribution model parameter values were: tissue conductivity 0.2
S/m (see below), tissue dielectric constant 3x106 (chosen to obtain a good fit), and electrode capacitance 0.58 µF.
3.4 Discussion
Voltages generated by the DBS electrode were recorded in vitro and in vivo and compared to solutions of volume conductor finite element models. The results demonstrate that appropriately parameterized models can accurately represent both the spatial and temporal characteristics of voltage fields induced by DBS. In turn, this study validates the field component of the field-neuron models used to estimate the neural response to applied stimulation (Miocinovic et al., 2006; Butson et al., 2007). Tables 3-1
82
Figure 3-6. Stimulation waveforms from in vivo and in vitro recordings, and their corresponding model solutions. The right and left columns shows the time course of waveforms generated by voltage-controlled and current-controlled stimulation. In both cases, models can accurately predict the temporal characteristics of the voltage recorded in the medium. All waveform were normalized so that peak cathodic amplitude equals -1.
83 and 3-2 provide a summary of the optimized parameters for modeling DBS-induced
voltage fields, as a static FEM or time varying Fourier FEM.
Electrode interface parameters define electrode capacitance according to the
equation:
C=εr*ε0*A/d
-12 2 where εr is interface relative permittivity, ε0 is permittivity of free space (8.85*10 C /
N m2), A is interface surface area (i.e. area of the electrode contact), and d is interface
thickness. In our model, value for d was set to 0.1 mm and εr calculated so that electrode
capacitance equals 0.6 µF. Capacitance of human DBS electrode (Medtronic 3387/3389)
has been estimated at 3.3 µF (Butson and McIntyre, 2005) and its surface area is 5.1
times larger than the monkey DBS electrode used in this study. Since the electrodes are
made from the same platinum/iridium alloy, their capacitance scales linearly with surface
area (3.3 µF / 5.1 = 0.64 µF). As a result, the same relative permittivity constant can be used for both human and monkey electrode model. However, explicit confirmation that
the other parameters are appropriate for modeling of human DBS leads remains to be
established.
Table 3-1. Parameters optimized for modeling DBS-induced voltage fields in saline and subcortical gray matter – static model Saline (0.9% NaCl) Tissue medium conductivity (S/m) 1.5 0.2 interface or encapsulation 0.37 0.015 conductivity (S/m) interface thickness or 0.1 0.25 encapsulation (mm)
84 Table 3-2. Parameters optimized for modeling DBS-induced voltage fields in saline and subcortical gray matter – Fourier model Saline (0.9% NaCl) Tissue medium conductivity (S/m) 1.5 0.2 medium relative 50 3*106 permittivity electrode interface relative 5.8*106 5.8*106 permittivity electrode interface thickness 0.1 0.1 (mm) interface or encapsulation 0.37 0.015 conductivity (S/m) encapsulation thickness n/a 0.25 (mm)
Electric field models commonly assume an ideal electrode behavior (i.e. no voltage drop across the metal-tissue interface). However, in both the in vitro and in vivo situation we found that recorded voltages were lower than the ideal model predictions. It was therefore necessary to include a resistive interface layer which was modeled as thin film approximation. By varying either conductivity or thickness of this layer it was possible to approximate a voltage drop resembling that recorded experimentally. In the tissue model, the encapsulation layer provided the resistive component, so explicit interface layer was not necessary.
The encapsulation sheath forms around the chronically implanted electrode because of the body’s inflammatory response to the foreign object. This collection of cellular infiltrate, protein deposits, and collagen matrices increases electrode impedance and reduces the resulting electric field in the tissue. As a result, the effective strength of voltage-controlled stimulation is reduced because the injected current is inversely proportional to electrode impedance. Interestingly, stimulation itself reduces electrode impedance, but the effect is transient and impedance increases once stimulation is
85 stopped. These effects have been studied in cochlear (Ni et al., 1992; Newbold et al.,
2004; Shepherd and McCreery, 2006), cortical (Weiland and Anderson, 2000),
subcutaneous (Grill and Mortimer, 1994), and DBS (Hemm et al., 2004) electrodes.
Impedance reduction by electrical stimulation has also been used to improve signal to
noise ratio of chronically implanted cortical recording electrodes (Johnson et al., 2005).
In our experience, most of the reduction in impedance occurred within the first 10-15 minutes of acute stimulation although the results were variable.
A limited number of histological studies have examined the encapsulation response around clinical DBS electrodes (Caparros-Lefebvre et al., 1994; Haberler et al.,
2000; Moss et al., 2004). In most cases mild gliosis has been observed around the electrode track (extending ~1 mm) and no overt neuronal damage in response to chronic
stimulation was found. Animal used in this study underwent acute stimulation only, but
our preliminary results suggest that chronic impedance values would be similar to those
recorded after ~1 hour of acute stimulation. However, it should be noted that this animal was receiving chronic steroid treatment for intestinal disorder so its inflammatory response to the DBS electrode could have been reduced.
There exist several possible sources of error in experimental voltage field recordings; most notably inaccuracies in the distance estimate between stimulating and recording electrodes, and DBS and microelectrode impedance variations. Recorded voltages could be under- or over-estimated because of recording distance uncertainty.
This is especially true for very small distances where voltage field changes rapidly. This issue could be minimized for in vitro recordings by direct visual inspection of electrode positions and repeated measurements under the same conditions. For in vivo recordings it
86 was more difficult to control the experimental conditions. Both microdrive coordinates
and x-ray images were used to estimate the distance between the electrodes. If the x-ray
image was not taken in the same plane defined by the electrodes, the distance between
them would appear smaller. While every effort has been made to minimize possible
errors (see Methods), submillimeter inaccuracies were possible. Furthermore, the DBS
electrode is flexible so once implanted and anchored to the head chamber, it exhibited slight curvature along its length. As a result, the DBS electrode and microelectrode may not have been perfectly parallel. This could be compensated for because both individual
DBS electrode contacts and microelectrode shaft could be visualized, and their distance measured. The microelectrode tip could not be reliably identified in the x-ray so its location with respect to the active DBS contact was determined from the depth were the largest voltage was recorded (e.g. at curve peak in Fig. 3-3, the microelectrode tip is closest to the active DBS contact).
Both DBS electrode impedance and microelectrode impedance affect the magnitude of the recorded voltage field. Variations in the in vivo DBS electrode impedance were observed as a function of time from electrode implantation and as a function of stimulation history. Acute stimulation reduced electrode impedance, and the changes started occurring within seconds of the stimulation and persisted for an hour or more. The greatest drop occurred within the first 10 minutes however this time was variable. After several hours of stimulation and voltage field recording, the electrode impedance was similar to the impedance recorded immediately post-implantation. After several days without stimulation, the impedance was again increased (compared to the post-stimulation levels). While the trends were clear, the precise impedance values could
87 not be controlled so voltage fields were recorded under variable conditions. We therefore separated data into groups according to post-stimulation impedance values for that day.
Interestingly, the ‘low impedance’ day started off with a fairly large value (12.5 kΩ) so the first track of that day (at 2.2 mm from DBS electrode) overlapped with ‘high impedance’ data.
Microelectrode impedance will also affect the amplitude of the recorded response; a decrease in the impedance will increase the response. The impedance will depend on the surface area of the exposed tip which can be increased by passing current through the electrode (‘bubbling’). A tip area that is too large will however reduce the spatial resolution of the recordings. We therefore used electrodes with ~0.5 MΩ impedance.
Point source field recordings confirmed that appropriate voltages values could be recorded with such microelectrodes (data not shown). Impedances were monitored during the experiments (before and after recordings) and once they deviated from the original value by more than +/- 0.2MOhm they were replaced.
The stimulation amplitude (0.3 V) was chosen to be low enough so that it did not induce any behavioral response in the animal. However, it should be noted that the stimulation probably excited some neurons in very close proximity to the electrode.
These stimulation-induced potentials, as well as spontaneous neuronal potentials, could contribute to the recorded voltage in the tissue medium, but these effects are probably negligible since neurons induce only microvolt extracellular potentials. In an attempt to alleviate these effects, we concentrated on voltage recordings from the beginning of the stimulus pulse and used low frequency stimulation.
88 3.5 Conclusion
Computational volume conductor models are commonly used to estimate neuronal response to electrical stimulation. However, prior to this study direct validation measurements of the voltage distribution generated by stimulation of subcortical gray matter structures have not been performed. We measured voltages generated by DBS electrodes in a saline bath, and in the thalamus of a monkey. We have shown that relatively simple finite element models can be used to accurately capture both spatial and temporal properties of the induced potentials.
89 Chapter 4: Computational analysis of subthalamic nucleus and
lenticular fasciculus activation during therapeutic deep brain
stimulation
Miocinovic S, Parent M, Butson CR, Hahn PJ, Russo GS, Vitek JL, McIntyre CC.
Computational analysis of subthalamic nucleus and lenticular fasciculus activation during therapeutic deep brain stimulation. J Neurophysiol. 2006 Sep; 96(3):1569-80.
4.1 Introduction
The subthalamic nucleus (STN) represents the most common anatomical target for DBS treatment of PD (Limousin et al. 1998). Electrodes placed in the STN are surrounded by several neural types (local projection neurons and their axons, fibers of passage, afferent inputs, etc.), but knowledge of the response properties of these different neural types to DBS is limited. Therapeutic DBS electrode contacts are typically located in the region of the dorsal STN, lenticular fasciculus (LF or Forel’s field H2) and zona incerta (Voges et al. 2002; Saint-Cyr et al. 2002; Starr et al. 2002; Hamel et al. 2003;
Yelnik et al. 2003; Zonenshayn et al. 2004; Nowinski et al. 2005). STN projection neurons send their highly collateralized axons to the globus pallidus, striatum and substantia nigra (Sato et al. 2000). The LF courses just dorsal to the STN and is composed of fibers from the internal segment of the globus pallidus (GPi), which carry output from the basal ganglia to the thalamus (Parent et al. 2001; Parent and Parent
2004). Given that GPi DBS provides similar therapeutic benefits as STN DBS (Burchiel et al. 1999; Obeso et al. 2001; Rodriguez-Oroz et al. 2005), both STN projection neurons
90 and pallidothalamic (GPi) fibers represent viable candidates as the therapeutic target of
the stimulation (Parent and Parent 2004). However, the axons of local projection neurons
and fibers of passage respond at similar extracellular stimulation thresholds (Ranck 1975;
Nowak and Bullier 1998; McIntyre and Grill 1999) making it difficult to determine which neuron types are activated during STN DBS.
The extent of neural activation generated by extracellular stimulation depends on the stimulation parameters, electrode and tissue electrical properties, and the position and orientation of neural elements with respect to the electrode (Ranck 1975; Tehovnik 1996;
McIntyre et al. 2004a). To address these issues in the context of STN DBS, we developed a comprehensive computer model with detailed representation of the 3D neuroanatomy, the time-dependent electric field generated by DBS electrodes, and the underlying biophysics that regulate the neural response to stimulation. We tested our hypothesis that both STN projection neurons and GPi fibers of passage are activated during clinically effective STN DBS. This would imply that both neural types could play a role in the therapeutic mechanisms of STN DBS, and prompt further investigation into electrode localization and stimulation parameter selection techniques for optimizing the stimulation to individual subjects.
We customized our modeling framework to analyze neural activation in two parkinsonian macaques implanted with chronic scaled clinical DBS systems (Hashimoto et al. 2003). Our goal was to theoretically reproduce the experimental effects of STN
DBS that improved parkinsonian symptoms (bradykinesia and rigidity) in the two monkeys. Our first aim was to quantify the proportion of STN projection neurons and
GPi fibers of passage that were activated during clinically effective and ineffective
91 stimulation. We found a significant increase in axonal activation of STN projection
neurons during clinically effective compared to ineffective stimulation. Considerable GPi
fiber activation was observed in only one of the two monkeys. Single unit extracellular
recordings of short latency, presumably antidromic, GPi activation during STN DBS in
both monkeys support our model predictions. The second aim was to analyze how
changes in electrode location affected neural activation. We found that sub-millimeter shifts can alter neural activation, highlighting the importance of precise electrode
positioning. The final aim was to investigate the effects of stimulation-induced trans-
synaptic inhibition on somatic and axonal firing in STN projection neurons during STN
DBS. We found that somatic firing was reduced during DBS compared to the
spontaneous pre-stimulation activity, as seen experimentally (Welter et al. 2004; Filali et
al. 2004; Meissner et al. 2005). However, axonal activation was largely unaffected by the
somatic inhibition and the axon easily fired at the stimulation frequency (McIntyre et al.
2004a).
4.2 Methods
The implementation of chronic scaled clinical DBS systems in parkinsonian
macaques provides the foundation for detailed study of the therapeutic mechanisms of the
stimulation (Hashimoto et al., 2003; Elder et al. 2005). We coupled our experimental
results with detailed computer modeling to provide new insight into the cellular effects of
STN DBS. We developed computational models customized to two parkinsonian
macaques implanted with STN DBS systems. Each model consisted of three fundamental
components: 1) a 3D anatomical model of the monkey basal ganglia, 2) a finite element
92 model of the DBS electrode and electric field transmitted to the tissue medium, and 3) multi-compartment biophysical models of reconstructed STN projection neurons, GPi fibers of passage and internal capsule fibers of passage. Populations of the neuron models
were placed within context of the 3D anatomical model and the histologically defined
DBS electrode positions. The DBS electric field model was then applied to the neuron
models, thereby allowing for theoretical prediction of the neural response to the
stimulation.
4.2.1 Brain Atlas
We built a three-dimensional (3D) reconstruction of the basal ganglia for monkey
R7160 (Macaca mulatta) (Hashimoto et al., 2003). Digital atlas templates of the
macaque brain (Martin and Bowden 2000) were warped in Edgewarp v3.28 (Bookstein
1990) to histological brain slices to identify borders of nuclei not visible directly in the
Nissl stained sections (Fig. 4-1). Nuclei of interest were outlined in the 2D warped atlas
slices, spaced in 1 mm increments. 3D volumes were created by interpolating between
these contour lines using the graphical modeling program Rhinoceros v3.0 (McNeal &
Associates, Seattle, WA). The resulting 3D brain atlas provided an anatomically realistic
virtual space to position the DBS electrode and the neuron models (Fig. 4-1). The DBS
electrode trajectory was reconstructed from the histological slices, and it was verified that it matched the electrode position drawing in the top row of Figure 1 from Hashimoto et
al. (2003). The brain of the second monkey (R370) was not available; therefore, we used
the same 3D atlas, but manually adjusted the DBS electrode position until the sagittal
93
Figure 4-1. Three-dimensional (3D) reconstruction of the basal ganglia for monkey R7160. Atlas templates (top right, Martin and Bowden 2000) were warped to Nissl- stained brain slices (top left) using Edgewarp (bottom, Bookstein 1990). Bottom right image shows a warped atlas template that now matches the histological slice, allowing definition of the nuclear borders. Outlines of nuclei from a series of warped atlas templates were interpolated into 3D volumes (right). A 4-contact deep brain stimulation (DBS) electrode was positioned in the posterior subthalamic nucleus (STN) of the customized brain atlas based on the histologically defined electrode location. Each leg of the 3D scale bar is 5 mm (D, dorsal; P, posterior; M, medial). Put, putamen; Caud, caudate nucleus; GPe, globus pallidus pars externa; GPi, globus pallidus pars interna; OT, optic tract; STN, subthalamic nucleus; Th, thalamus including thalamic reticular nucleus.
94
Figure 4-2. Multicompartment cable model of an STN projection neuron. A: 3D reconstruction of a macaque STN neuron generated in Neurolucida. B: neuron geometry was imported into the NEURON simulation environment. Soma and neuronal processes were divided into compartments and modeled as a series of resistors and capacitors. C: membrane lipid bilayer was represented as a capacitor and ion channels as variable resistors. Na, fast sodium channel; NaP, persistent sodium channel; KDR, potassium delayed rectifier; Kv31, potassium fast rectifier; sKCa, small conductance calcium activated potassium channel; Ih, hyperpolarization-activated cation channel; CaT, low voltage–activated calcium channel; CaN and CaL, high-voltage activated calcium channel; L, leak channel; Cm, membrane capacitance (see Gillies and Willshaw 2006 for details on the neuron membrane dynamics).
cross-section from the 3D atlas matched the rendering in the bottom row of Figure 1 from
Hashimoto et al. (2003).
4.2.2 Neuron Geometries
The 3D geometry of a longtailed macaque’s STN projection neuron (Macaca fascicularis) was reconstructed using biotin dextran amine labeling and axonal tracing as described by Sato et al. (2000) (Fig. 4-2). The neuron geometry was defined using
95 Neurolucida (MicroBrightField, Inc., Williston, VT) and converted for display in
Rhinoceros and our 3D brain atlas. Axonal tracing experiments (Sato et al. 2000) have
revealed that STN projection neurons course either dorsally along the ventral border of
the thalamus or ventrally along the lateral border of the STN on their way to the globus
pallidus. To account for this anatomical variability, the original neuron reconstruction was used to create two additional STN neuron geometries with alternative axonal paths
(Fig. 4-3A). The GPi axon geometry was based on the description of lenticular fasciculus trajectory from Parent and Parent (2004). The LF fibers emerged dorsomedially from the
GPi, crossed the IC at a level dorsal to the STN, turned caudally to run along the dorsal
border of the STN in Forel’s field H2, joined the ansa lenticularis in Forel’s field H,
continued through Forel’s field H1, and terminated in the ventral thalamus (Fig. 4-3B).
The internal capsule defines the lateral border of the STN. Consequently, motor
evoked responses from activation of the corticospinal tract (CST) can be elicited with
relatively low thresholds during STN stimulation. To provide a gross level of model
validation and connection to behavioral measurements we incorporated CST axon
trajectories into our anatomical framework. From the level of dorsal thalamus, the CST
fibers coursed ventrally at an approximately 20 degree anterior-to-posterior angle (Fig. 4-
3C).
The three STN neuron geometries, along with the GPi and CST axon trajectories were placed within the 3D atlas in their respective anatomically realistic positions and orientations (Fig. 4-3). Neural populations of the STN neurons, GPi fibers and CST fibers were created by copying the five basic geometries and distributing them randomly within the atlas, while still keeping each within their respective anatomical boundaries. In total,
96
Figure 4-3. Neural populations and DBS electrode in the context of 3D neuroanatomy. A: 3 types of STN projection neurons share the same soma and dendritic tree but have different axonal trajectories. B: GPi fibers form the lenticular fasciculus on their way to the ventral thalamus. C: population of internal capsule fibers. Diameters of neuronal processes were increased 10 times for figure rendering. Each leg of 3D scale bar represents 1 mm (D, dorsal; P, posterior; M, medial).
97 three such populations were created by randomly shifting neurons by ± 250 µm in any direction, and manually repositioning those that ended up outside their anatomical boundaries. After the populations were established, a DBS electrode was placed within the STN. The cells were defined as damaged if any part of the axon, soma or substantial portion of the dendritic tree intersected with the electrode. We started with 100 STN neurons, 80 GPi fibers and 70 CST fibers, removed all the damaged cells and then randomly removed additional cells so that the final count for each of the three populations was 50 cells.
4.2.3 Neuron Biophysics
We built multi-compartment cable models of STN projection neurons, GPi fibers of passage and CST fibers of passage using NEURON v5.8 (Hines and Carnevale 1997).
The STN soma, initial segment and dendrites contained channel dynamics of the rat subthalamic projection neuron originally developed by Gillies and Willshaw (2006), where dendritic channel conductance densities scale linearly with distance from the soma
(Fig. 4-2C). To better fit the membrane dynamics to our neuron geometry and the firing pattern of STN neurons in a parkinsonian monkey, we modified the original conductances in the following way: the calcium-activated potassium channel was increased by 80%, the fast potassium rectifier was increased by 20% (soma and initial segment only), and the fast acting sodium channel was decreased by 25% in the soma and initial segment, and by 35% in the dendrites. The model temperature was set to 36°C to simulate in-vivo conditions. The STN neuron model had a resting potential of
approximately -54 mV and spontaneous tonic firing of 32 Hz consistent with the rate of
98 36.5±10.8 Hz recorded by Wichmann et al (2002) in the parkinsonian macaque. The
neuron firing rate increased with increasing amplitude of intracellular depolarizing
current. The slope of the model frequency-intensity (f-I) curve was 0.26 Hz/pA, lower than slopes recorded in rat brain slices (~0.54 Hz/pA in Bevan and Wilson 1999). Despite
this discrepancy the model neuron was able to fire at more than 200 Hz, well above the
range of clinically used DBS frequencies (100-185 Hz). The input resistance measured as
the slope of the I-V curve while neuron was in a hyperpolarized state (-59.6 mV resting
potential) and injected with small currents (-0.2 nA to 0.05 nA) was 35 MΩ, twice the
value of resistances recorded in rat STN neurons in-vivo (mean 18 MΩ, range 9-28 MΩ
in Kita et al. 1983). This discrepancy could be due to a lack of synaptic inputs into the
dendritic tree, differences between monkey and rat STN neuron morphology, or
imperfect measurement of dendritic diameters during histological 3D reconstruction
resulting in underestimate of the surface area. Average rat in-vitro measurements range
from 146-200 MΩ (Nakanishi et al. 1987; Beurrier et al. 1999) placing the model neuron
within the range of in-vivo and in-vitro recordings. The membrane time constant
estimated from the 1/e point of the membrane potential change induced by a low intensity hyperpolarizing current pulse was 6 ms consistent with in-vivo rat measurements (6±2 ms
in Kita et al. 1983).
The axon of the STN neuron as well as the GPi and CST axons were based on the
myelinated axon model originally described by McIntyre et al. (2002). The fiber diameter
was set to 2 µm and the individual segment dimensions and ion channel conductances
were adjusted as previously described (McIntyre et al. 2004a). The axonal resting
potential was set to -65 mV. The GPi axon was induced to fire tonically at 80 Hz by
99 injecting short current pulses at the GPi terminal end of the axon to mimic parkinsonian
macaque GPi firing rate of 80.1±20.2 Hz (Wichmann et al. 2002). There were no synaptic
connections or interactions between any of the different neurons in the model system.
In some simulations trans-synaptic GABAa input to the STN neurons was added
to evaluate the influence of stimulation-induced trans-synaptic conductances on somatic
activity during high frequency stimulation. This was simplistically modeled as an
inhibitory postsynaptic current applied only to the central compartment of the cell body.
The time course and amplitude of the GABAa synaptic conductance was modeled with
experimentally defined first-order kinetics of the transmitter binding to postsynaptic
receptors (Destexhe et al. 1994a,b). Since afferent inputs (i.e. axonal terminals) are
typically more excitable than the passing axons (McIntyre et al. 2004a; Anderson et al.
2006), in all applicable simulations we assumed that synaptic inhibition was always
activated by each extracellular stimulus pulse, regardless of the cell position with respect to the electrode.
4.2.4 Electric Field Model
An axisymmetric finite element model (FEM) of the DBS electrode was created in FEMLAB 3.1 (COMSOL, Inc.,Burlington, MA) to calculate potentials generated in the
tissue medium by the electrode. The stimulating lead was a scaled-down version of the
chronic DBS electrode used in humans (Model 3387, Medtronic Inc., Minneapolis, MN)
and consisted of four metal contacts each with a diameter of 0.75 mm, height of 0.50 mm,
and separation between contacts of 0.50 mm. The volume conductor was 5 cm x 5 cm in
size and grounded at the boundaries. The bulk conductivity of the tissue medium was 0.2
100 S/m (Ranck 1963), and a 0.25 mm encapsulation sheath (0.1 S/m) surrounded the electrode shaft (Grill and Mortimer 1994; Butson et al. 2006). The electrode shaft was
modeled as an insulator (1e-6 S/m) and each electrode contact as a conductor (1e6 S/m).
Voltage sources were specified at two electrode contacts for bipolar stimulation. The
model had variable resolution mesh with a total of 32,640 mesh elements that increased
in size with increasing distance from the electrode. Voltage values within the volume
were determined from the Poisson equation, which was solved using direct matrix
inversion (UMFPACK solver) (Fig. 4-4).
The stimulus waveform produced by the Itrell II (Medtronic Inc., Minneapolis,
MN) implantable pulse generator (IPG), as used in human DBS and the monkey
experiments, is a voltage-controlled biphasic asymmetric square pulse. However, the
actual stimulus delivered to the brain tissue is modified by the electrode capacitance. To
account for the effects of electrode capacitance, a Fourier finite element model (FEM)
was utilized as described in Butson and McIntyre (2005). The four general steps of this
method are briefly described here. First, the IPG stimulus waveform was constructed in
the time domain. It was then converted to frequency domain using discrete Fourier
transform (DFT) in Matlab (Mathworks, Natick, MA). Third, the FEM model was solved
at each component frequency of the DFT (1024 frequencies between 0 and 50kHz). The
result at each frequency was scaled and phase shifted using the DFT magnitudes and
phases. Finally, an inverse Fourier transform was performed to obtain the stimulus
waveform in the time domain. The electrode was modeled as purely capacitive (0.65
µF), and adjusted to account for the smaller surface area of monkey electrode contacts
compared to the human DBS electrode (Butson and McIntyre 2005).
101
Figure 4-4. Field-neuron model of STN DBS. A: finite element model (FEM) voltage solution for 1-V bipolar stimulus overlaid with a sagittal brain cross-section from monkey R7160. A 250-µm-thick encapsulation layer surrounds the DBS electrode. Str, striatum; GPe, globus pallidus pars externa; GPi, globus pallidus pars interna; OT, optic tract; Th, thalamus including reticular thalamic nucleus; STN, subthalamic nucleus; Sn, substantia nigra. B: extracellular potentials generated by the electrode create transmembrane polarization along the STN projection neuron. Neural compartments are colored according to their transmembrane potential at onset of a subthreshold stimulus pulse. Arrows indicate depolarized nodes of Ranvier. Diameters of neuronal processes were thickened for figure rendering.
4.2.5 Stimulation Prediction
We coupled the finite element electric field model with the multi-compartment neuron models to enable quantitative stimulation predictions in the context of the 3D neuroanatomy. This coupling was accomplished by applying extracellular voltages from the electric field model to each compartment of each neuron model and simulating the biophysical response (action potential signaling) of each neuron over time (McIntyre et al. 2004a) (Figs. 4-4, 4-5). The magnitude of the applied extracellular voltage was
102 dependent on the stimulus amplitude, stimulus waveform, and the compartment’s
distance from the electrode. At each time step of the simulation the extracellular voltage
at each neural compartment was updated to a value determined by the time-dependent
stimulus train delivered to the tissue medium (Figs. 4-4, 4-5).
Clinical efficacy for various DBS parameter settings was established in two
parkinsonian macaques using behavioral tests (for details see Hashimoto et al. 2003). In
both monkeys, R7160 and R370, the electrode was positioned in the posterior STN at a
20 degree angle in the sagittal plane. In monkey R7160, bipolar stimulation (contact 0
cathode, contact 2 anode) at 136 Hz and 210 µs pulse width produced consistent
improvement in rigidity and bradykinesia at 1.8 V amplitude (clinically effective
stimulation), but not at 1.4 V (clinically ineffective stimulation). In monkey R370 bipolar
stimulation (contact 2 cathode, contact 0 anode) at 136 Hz and 90 µs pulse width was clinically effective at 3 V amplitude and clinically ineffective at 2 V. Thus, these various stimulation parameters were applied to our model system. The fundamental differences between the model simulations for the two monkeys were the electrode position and stimulation parameters. In addition, the two electrode positions resulted in somewhat different neural populations because different neurons were ‘destroyed’ by the electrodes.
The model neurons were stimulated with a train of 25 pulses. Longer train durations (1 second; 136 pulses) did not impact neural response to stimulation (Fig. 4-5B).
We did not observe any model neurons that exhibited a blocking of axonal firing from the direct application of the DBS electric field; therefore, we concentrated our analysis on the excitatory response of the stimulation (Fig. 4-5A). Those that produced orthodromically propagating action potentials in response to more than 80% of the
103 stimulus pulses were considered to be activated. This percentage was chosen because most neurons responded either to none or to 20 or more of the 25 stimulus pulses. Some activated neurons did not respond to all 25 pulses because in certain cases the stimulus pulse was delivered immediately following a spontaneous action potential (originating in the soma), while the axon was still in the refractory state. Percentages of activated neurons were averaged over three randomized populations. Student’s t-test (one-tailed; p<0.05) were performed to compare the averages for clinically ineffective and effective stimulation conditions.
4.2.6 Experimental Recordings of GPi Activity During STN DBS
The experimental recording procedure and data collection from the two monkeys used in this study has been described in detail elsewhere (Hashimoto et al., 2002; 2003;
Elder et al., 2005). Briefly, single unit neural activity was recorded extracellularly from
GPi identified neurons using glass-coated platinum–iridium microelectrodes (impedances
of 0.4–0.8 MΩ at 2 kHz). Recording penetrations were made in parasagittal planes
moving rostral to caudal at an angle of 70° to the orbitomeatal line. Neurons (n = 12 for
monkey R7160 and n=27 for monkey R370) were recoded for 25-35 sec during clinically
effective STN DBS at 136 Hz. Stimulus artifact template subtraction methods
(Hashimoto et al., 2002) and in-house software developed in MATLAB v7.0 (Mathworks
Inc., Natick, MA) were used for the neural signal analysis. Peristimulus time histograms
were constructed with a 0.2 ms bin size. The histograms used 336,373 spikes resulting
from 644,929 stimuli in 39 GPi cells. Stimuli for which no spike was recorded in the
interstimulus interval were not considered. Probability distributions for R7160 and R370
104
Figure 4-5. STN neuron firing in response to extracellular stimulation. Lowercase letters indicate location in the STN neuron where the transmembrane voltage was recorded. a, soma; b, 1st node of Ranvier; c, 30th node of Ranvier; d, 50th node of Ranvier. A: action potential initiated in the axon and propagated toward the cell body and axonal terminals in the globus pallidus. Traces in the top row represent stimulus voltage waveforms applied to the neuron. The 4 traces below show response to a subthreshold DBS pulse (1.4 V), a suprathreshold DBS pulse (1.8 V), and a suprathreshold DBS pulse (1.8 V) in a model with inhibitory somatic synapses. B: STN neuron spontaneous firing and response to stimulation are stable over time. The 4 traces show neuronal firing before, during, and after 1-s, 136-Hz DBS train in the soma and the distal axon of the same neuron. Results are displayed for models with and without GABAa stimulation induced synaptic input (0.013 µS). Somatic firing rate was lower than the stimulation frequency because action potentials initiated in the axon did not invade the soma for every stimulus pulse. GABAergic synaptic inputs reduce somatic firing, but axonal output was largely unaffected (see also Fig. 4-9).
105 were compared using a χ2 goodness of fit test (p<0.05). Cells were further classified as
having an early response if more than 15% of the stimulus responses occurred at a
latency of less than 1.5 ms. Significance of differences in the proportion of early response cells was found by χ2 test (p<0.05).
4.3 Results
We calculated levels of axonal activation for populations of STN projection neuron, GPi fibers of passage (lenticular fasciculus) and CST fibers of passage during
clinically effective and ineffective STN DBS in two parkinsonian macaques. We
evaluated four general aspects of our model system. First, activation of CST fibers at
muscle contraction thresholds were analyzed to provide a gross degree of model
validation. Second, we evaluated the neural response to therapeutic stimulation using three separate randomized populations of STN neurons and GPi fibers, and we correlated the model predictions to single-unit microelectrode recordings from the two monkeys during STN DBS. Third, the sensitivity of neural activation to electrode position was
assessed by moving the electrode by 0.25 mm in four directions in the horizontal plane.
And finally, the effects of DBS on STN somatic firing were evaluated in a model with stimulation-induced inhibitory trans-synaptic conductances.
4.3.1 Activation of the Internal Capsule with DBS
To address the experimental predictability of our model system we evaluated the
activation of CST fibers. Visually determined muscle contraction thresholds in the two
monkeys, were 3V for R7160 and 3.5V for R370. These stimulation parameters resulted
106 in activation of 11±1% and 9±3% of CST fibers in our R7160 and R370 models,
respectively, averaged over three random populations (mean±SD). In addition, clinically
effective stimulation parameters resulted in minimal activation of CST fibers, 0% and
5±2% in the R7160 and R370 models, respectively. These results indicate that the model exhibited appreciable increases in CST fiber activation when comparing stimuli that experimentally resulted in no visible muscle contraction (clinically effective) and stimuli where muscle contractions were observed (CST threshold). This provides some evidence that the voltage spread in the tissue around the electrode was accurately predicted by the finite element model, and that the neuron models fire at realistic activation thresholds.
4.3.2 Activation of the STN and LF with STN DBS
In the context of STN DBS, both STN projection neurons and GPi fibers of passage in the LF represent viable candidates as the therapeutic target of the stimulation.
The GPi is an output nucleus of the basal ganglia and the STN modulates basal ganglia
output through excitatory projections into the GPi. We defined DBS induced activation of
STN projection neurons as the generation of an action potential anywhere in the neuron,
resulting from the applied electric field, which propagated to the axonal terminal. When
stimulating STN projection neurons with DBS, action potential initiation always took
place in the myelinated axon, and if an action potential was induced by the stimulation it
always propagated to the axon terminal (Fig. 4-5A).
Our simulations predict activation of both GPi fibers and STN neurons during
STN DBS (Fig. 4-6). These results were consistent over three randomized populations of
neurons. Stimulation parameters that failed to improve parkinsonian symptoms in
107 monkey R7160 (1.4 V, 210 µs, 136 Hz), activated 29±2% of STN projection neurons,
9±4% of GPi fibers of passage and no CST fibers. Clinically effective stimulation (1.8 V) activated 37±4% of STN projection neurons (30% average increase), 18±6% of GPi fibers of passage (107% increase) and no CST fibers (Fig. 4-6A). In the second monkey,
R370, clinically ineffective (2 V, 90 µs, 136 Hz) and clinically effective stimulation (3 V,
90 µs, 136 Hz) activated 31±3% and 49±5% of STN neurons (54% increase), 66±2% and
82±6% of GPi fibers (24% increase) and 1±1% and 5±2% of CST fibers, respectively
(Fig. 4-6B). The increase in activation between clinically ineffective and effective stimulation was statistically significant for STN neurons in both monkeys and for GPi and CST fibers in monkey R370 (p<0.05) (Fig. 4-6).
The therapeutic effects of DBS depend critically on the stimulation frequency, with frequencies over 100 Hz being generally beneficial and those below 50 Hz sometimes worsening the symptoms (Rizzone et al. 2001). However, the output of our model system was not substantially affected by the stimulation frequency (i.e. approximately the same numbers of neurons were activated by the stimulus train). We did observe small decreases in the number of neurons activated as the frequency decreased from 136 Hz to 2 Hz. For example, monkey R7160 exhibited an 8% reduction in STN neurons and a 2% reduction in GPi fibers activated when comparing 136 Hz and 2 Hz stimulation. More importantly, all neurons activated by a given stimulus pulse fired in synchrony, and as the stimulation frequency changed the activated neurons were entrained to fire at the given stimulation frequency. These results suggest that changes in stimulation frequency do not dramatically affect the volume of tissue activated, but rather that stimulation frequency exerts its influence on the basal ganglia network functioning
108
Figure 4-6. Neural activation during clinically effective and ineffective DBS. A: STN neuron axons (top) and GPi fibers (bottom) activated by clinically effective DBS in monkey R7160 (1.8 V, 136 Hz, 210 µs, contact 0 cathode, contact 2 anode) are shown in red. Axons that did not respond to >80% of the stimulus pulses are shown in gray. Each leg of the 3D scale bar represents 1 mm. B: percent of neurons activated for monkeys R7160 (top) and R370 (bottom) during clinically ineffective, clinically effective, and corticospinal tract threshold DBS, averaged over 3 randomized populations. Only corticospinal tract (CST) fiber activation was evaluated at CST threshold amplitudes. Asterisks indicate significant difference between clinically ineffective and effective stimulation (P < 0.05; t-test).
(Montgomery and Baker 2000; Rubin and Terman 2004; Grill et al. 2004), which was not addressed in this model.
Given the limited quantitative characterization of STN neuron membrane dynamics, we performed a range of sensitivity analyses to addresses the robustness of our
DBS model predictions. Before finalizing the variant of the Gillies and Willshaw (2006) membrane dynamics used in this study we evaluated a wide range of models of STN
109 neuron membrane dynamics (Wilson et al. 2004; Otsuka et al. 2004). The model used in
this study best matched the available in vitro experimental characterization of STN
neuron firing. However, nearly every STN neuron model variant that we evaluated
generated nearly identical neural activation results in response to STN DBS. This consistency in the model output was primarily linked to the fact that action potential initiation in response to extracellular stimulation takes place in the myelinated axon
(Nowak and Bullier, 1998; McIntyre and Grill, 1999; McIntyre et al., 2004a). In turn, the description of the somatic and dendritic membrane dynamics had little effect on the axonal output generated by DBS. For example, we modified the m (activation) gate of the spike triggering fast sodium channel in the dendrites, soma and initial segment. Large changes in the time constant (± 50%) and small shifts in the steady-state activation voltage (± 1mV) had no effect on the number of STN neurons activated by DBS (data not
shown). However, it should be noted that modifications of the steady-state activation
voltage strongly affected the spontaneous firing frequency of the STN neurons.
4.3.3 Experimentally Recorded Short-Latency GPi Activity During STN DBS
The response of GPi neurons to STN DBS was monitored experimentally by
single-unit microelectrode recordings. Utilizing stimulus artifact template subtraction we were able to construct peristimulus time histograms and analyze activity of GPi neurons immediately following each DBS pulse (Fig. 4-7) (Hashimoto et al., 2002; 2003).
Experimentally the activation of GPi fibers resulted in short latency excitation of GPi cell bodies which could be interpreted as antidromic activation (Bar-Gad et al., 2004). Using the 3D anatomical brain atlas and GPi axon model we estimated the propagation latency
110 from DBS activation of the LF to antidromic activation of the GPi to be < 1.5 ms. We calculated the number of experimentally recorded GPi cells that exhibited short-latency excitation for 15% or more of the DBS pulses, indicative of reliable antidromic activation
(see Discussion). In monkey R370 41% of GPi cells fired within 1.5 ms following a DBS pulse whereas only 8% of GPi cells fired within the same time period in monkey R7160
(Fig. 4-7). Consistent with these experimental results, our model predicted a large degree of GPi fiber activation for monkey R370 (82%), but a much smaller activation for monkey R7160 (18%).
Figure 4-7. Experimentally recorded GPi firing during STN DBS. Histograms of monkeys R7160 (A) and R370 (B) show the percentage of total GPi spikes (all spikes recorded from all cells in each animal) during clinically effective stimulation at 136 Hz separated into 0.2-ms bins over the interstimulus interval. Monkey R370 exhibited an increased number of spikes in the 1st 7 bins (<1.5 ms), indicating a higher degree of short latency excitation compared with monkey R7160.
111 4.3.4 Sensitivity of Neural Activation to Electrode Position
Electrode location can have a substantial impact on the therapeutic efficacy and side effects of DBS; however, conflicting opinions exist on the optimal position of STN
DBS electrodes (Voges et al. 2002; Saint-Cyr et al. 2002; Starr et al. 2002; Hamel et al.
2003; Yelnik et al. 2003; Zonenshayn et al. 2004; Nowinski et al. 2005). Therefore, we
examined how small modifications in electrode position affected our simulation results.
We compared neural activation for the original electrode position in monkey R7160 to
models with the electrode moved either 0.25 mm medial, lateral, anterior or posterior to
the original position (Fig. 4-8). For clinically ineffective stimulation, the range of
activation was 6-16% for GPi fibers, 0-2% for CST fibers and 12-30% for STN neurons.
For clinically effective stimulation, activation extended between 6-28% for GPi fibers,
between 0-10% for CST fibers and between 18-42% for STN neurons. As a result,
electrode location within the STN region can substantially affect axonal activation of
both STN neurons and GPi fibers.
4.3.5 STN Neuron Somatic Firing During STN DBS
In the simulations described above, STN neuron activation was assessed at the
distal end of the axon. Since experimental extracellular STN microelectrode recordings
register primarily somatic firing rather than the axonal output, we also evaluated firing frequency in the STN soma during STN DBS (Figs. 4-5 and 4-9). During extracellular stimulation, action potential initiation occurs in the axon, and as a result the soma and axon can fire independently (Nowak and Bullier 1998; McIntyre and Grill 1999;
McIntyre et al. 2004a). In addition, extracellular stimulation can activate synaptic inputs
112
Figure 4-8. Sensitivity of neural activation to electrode position. Percentage of STN neurons (top) and GPi fibers (bottom) activated during DBS for monkey R7160 with the electrode moved from the original position (O) to a location 0.25 mm medial (M), lateral (L), anterior (A), or posterior (P). The original electrode position was in the sagittal plane, 6.2 mm from the midline, at a 20° anterior-to-posterior angle. The center of the bottom contact (cathode) was 3.3 mm ventral to the AC-PC plane (horizontal plane defined by the anterior and posterior commissures) placing it in the posterior-medial- ventral border of the STN. The 2nd from the top contact (anode) was 1.4 mm below the AC-PC plane, corresponding to the area of LF just dorsal to posterior-lateral border of the STN.
impinging on local projection neurons (Baldissera et al. 1972; Gustafsson and Jankowska
1976; Dostrovsky et al. 2000; Welter et al. 2004; Filali et al. 2004; Meissner et al. 2005;
Anderson et al. 2006). Synaptic inputs to the STN come primarily from the external part
of globus pallidus (GPe) and cerebral cortex. Dominant GPe inhibitory afferents
converge primarily on the cell body and proximal dendrites (Smith et al. 1990), which we
modeled as GABAa trans-synaptic conductance in the central compartment of the soma.
113 The presence of stimulation induced GABAa synaptic input inhibited somatic firing and thus reduced the frequency of spontaneous spikes (Figs. 4-9A and 4-5B). The magnitude of somatic firing suppression induced by stimulated GABAa input depended on the neuron position relative to the electrode and the strength of the inhibitory synaptic conductance. In the absence of extracellular stimulation or inhibitory synaptic input, the output of the STN neuron was dictated by the rate of spontaneous somatic spiking (32
Hz). If the STN neuron was too far from the electrode to be directly activated by the extracellular stimulus, but high frequency inhibitory synaptic inputs were applied, the spontaneous somatic firing could be entirely suppressed, resulting in no axonal output.
However, if the neuron was close enough to the electrode to be directly activated by the extracellular stimulus, axonal firing (i.e. neural output) was almost completely dictated by the stimulation frequency and largely unaffected by inhibitory synaptic currents applied to the soma (Fig. 4-9B). Interestingly the inhibition of spontaneous somatic firing also increased the reliability of stimulus-evoked action potential generation in the axon
from extracellular stimulation because of the elimination of the possible interaction
between spontaneous spikes and stimulation induced spikes.
4.4 Discussion
The neural response to STN DBS that is responsible for the therapeutic effects of
the stimulation has been unclear. This limitation in scientific knowledge has hindered our
ability to understand and optimize this medical technology for current and future
applications. Using an anatomically and electrically detailed computer model, we evaluated the neural activation generated by clinically effective and ineffective DBS in
114
Figure 4-9. STN firing frequency under the influence of stimulation-induced trans- synaptic GABAa inhibitory inputs. Firing frequency was measured in the soma (A) and distal axon (B) of the same cell and averaged for a population of STN neurons (monkey R7160; clinically effective stimulation). Somatic firing, with respect to the no stimulation condition (No DBS), decreased with increasing strength of the GABAa synapse. Distal axon firing, with respect to the 0 conductance condition, was relatively unaffected by somatic inhibition.
two parkinsonian macaques. Our simulation results showed axonal activation of both
STN projection neurons and GPi fibers of passage during STN DBS. The relative proportion of neural types activated depended strongly on the position of the cathodic contact in the STN region. In monkey R7160, the cathode was in the ventral portion of the STN, which resulted in limited GPi fiber activation (~10-20%). In monkey R370, the cathode was in the dorsal STN, on the border with LF, resulting in greater GPi fiber recruitment (~65%) which also significantly increased during therapeutic stimulation conditions (~80%). These theoretical predictions were supported by the large number of
115 GPi neurons experimentally recorded with short latency excitation in monkey R370, indicative of LF antidromic activation, not seen in monkey R7160. Both monkeys had similar levels of STN neuron activation which showed significant increases with clinically effective stimulation (from ~30% to 40-50%). These results indicate that activation of approximately half of the STN is sufficient for the behavioral manifestation of the therapeutic effects of STN DBS. The additional recruitment of GPi fibers may also play an important role in therapeutic outcome, but large-scale activation of GPi fibers may not be necessary.
4.4.1 Model Development, Analysis, and Limitations
The computer model developed in this study utilized a number of anatomical, physiological, and electrical improvements over previous attempts to theoretically address the cellular effects of DBS (McIntyre et al. 2004a). However, when constructing such a comprehensive model it is necessary to make a number of assumptions and simplifications. The following section attempts to document these limitations and provide insight into their possible impact on our results.
We dedicated substantial effort toward the development of accurate neuron models for our simulations; however, they were unable to capture the full spectrum of experimentally defined characteristics. We utilized neuron biophysical properties based on rat STN neurons (Gillies and Willshaw 2006). Rat STN neurons have been extensively characterized with in-vitro preparations, making it possible to construct faithful model representations with Hodgkin-Huxley type channel dynamics (Terman et al. 2002;
Wilson et al. 2004; Otsuka et al. 2004; Gillies and Willshaw 2006). In addition, we
116 attempted to approximate in vivo conditions by adjusting both the STN projection neurons and GPi fibers of passage to match the spontaneous activity observed in parkinsonian macaques (Wichmann et al. 2002). Nonetheless, our STN DBS model exhibited a simple on/off activation outcome (Fig. 4-5B), whereas the therapeutic effects of DBS typically evolve over seconds, minutes, and even hours of stimulation (Temperli et al. 2003). This discrepancy in the model predictions and clinical observations may be related to our inability to fully characterize the STN neuron membrane dynamics. Recent in vitro experimental studies have shown slow inactivation of sodium channels may play an important role in STN somatic activity during high frequency stimulation (Beurrier et al. 2001; Do and Bean 2003). In addition, there exists a long list of factors influencing neural plasticity that could be affected by DBS (e.g. gene expression changes, channel density changes, synaptic strength changes, etc.) that were not included in our model system. However, our previous experience suggests that the specific details of the somatic ion channel membrane dynamics are of limited importance in the global effects predicted by computer models of extracellular stimulation (see sensitivity analyses in
McIntyre and Grill 1999, 2000; McIntyre et al. 2004a). Possibly of greater importance is the use of realistic neural morphologies which directly interact with the electric field and determine polarization profiles of each neuron (Figs. 4-4B and 4-5A). Although our model included a full 3D reconstruction of a macaque STN projection neuron (Sato et al.
2000), as well as GPi and CST axonal trajectories based on documented morphologies
(Parent et al. 2001), all neurons in each population were copies of a generic geometry. To compensate for this limitation we added diversity in our STN projection neurons by
117 creating two additional axonal trajectories based on histological axonal tracing
experiments (Sato et al. 2000).
The results suggest that electrode position with respect to the surrounding
anatomy is an important factor in the stimulation outcome (Figs. 4-6 and 4-8). Using histological brain slices we developed a 3D basal ganglia reconstruction for one of the
monkeys (R7160) and were able to precisely recreate the electrode position in the tissue.
Unfortunately, we did not have access to R370’s brain, so the same anatomical model
was used for both monkeys. However, the neuroanatomical differences are likely to be
small since the two monkeys were both rhesus macaques of similar size and the same sex.
Further, the electrode for R370 was positioned using the descriptions and drawings from
Hashimoto et al. (2003). It should also be noted that the three-dimensionally complex
tissue anisotropy around the STN can affect the neural response to DBS (McIntyre et al.
2004b). However, in this study we were unable to obtain diffusion tensor imaging data to
estimate the 3D tissue conductivity properties specific to these animals so an isotropic
conductivity was used for the bulk tissue medium. To address some of these issues, a
gross level model validation was attempted by comparing the experimentally defined
CST thresholds in the two monkeys with the corresponding model predictions (Fig. 4-6).
The models showed minimal CST activation during clinically effective stimulation (~0-
5% fibers activated), but more substantial activation at experimentally determined CST
thresholds (~10%). The exact percentage of CST fibers needed to be activated to generate
a noticeable muscle contraction is unknown. However, we believe that the increase in
CST activation demonstrated by the model is substantial enough to justify the assumption
that it would correspond to a perceptible effect experimentally. This is particularly true
118 for monkey R7160 whose CST activation went from zero to 11±1%, and whose brain we explicitly reconstructed in 3D and were able to determine electrode location with a high degree of certainty.
We evaluated somatic firing during high frequency stimulation (HFS) by including a highly simplistic representation of stimulation-induced trans-synaptic inhibition. STN microelectrode recording studies in monkey (Meissner et al. 2005) and human (Welter et al. 2004; Filali et al. 2004) have shown that STN HFS reduces the somatic firing rate of STN neurons, which our results support (Fig. 4-9A). By selecting the appropriate synaptic conductance value, our model is in close agreement with
Meissner et al. (2005) who documented an ~50% decrease in somatic firing frequency with STN HFS. Our simulations also show that axonal output is largely unaffected by the inhibition of somatic firing (Fig. 4-9B). This disconnect between axonal and somatic firing has previously been noted in a model of DBS of a thalamocortical relay neuron
(McIntyre et al. 2004a) and we now demonstrate it for a population of STN projection neurons.
It should also be noted that the STN is surrounded by many other fiber tracts that we ignored in this study. The reciprocal STN-GPe projections, STN-substantia nigra pars reticulata connections, and nigrostriatal fibers may also be directly affected by STN DBS
(Lee et al., 2006). While it is possible that these tracts play a role in therapeutic mechanisms of DBS, we limited the present study to evaluate the two most probable candidates (based on location of optimal therapeutic contacts in human patients), namely
STN projection neurons and pallidothalamic (GPi) fibers. We also ignored possible physiologic effects that could result from antidromic activation of afferent inputs
119 projecting to the STN (mostly from GPe and cerebral cortex), as well as the potential role
of glial cells in the regulation of the extracellular environment (e.g. extracellular ionic
concentrations and neurotransmitter levels). In turn, our model had a number of
limitations, but we do not think they impact our fundamental conclusions. In addition, the
model system used in this study represents one of the most anatomically and electrically
accurate computer models of DBS ever created and never before has such explicit connection between modeling and experimental DBS results been attempted.
4.4.2 Neural target of STN DBS
In current clinical practice, the STN is the target of choice for DBS treatment of
PD, even though GPi DBS is similarly effective (Burchiel et al. 1999; Obeso et al. 2001;
Rodriguez-Oroz et al. 2005). Several studies have shown that clinically effective STN
DBS electrode contacts are located in the dorsal STN, or the white matter above the STN formed by the LF and zona incerta (Voges et al. 2002; Saint-Cyr et al. 2002; Starr et al.
2002; Hamel et al. 2003; Yelnik et al. 2003; Zonenshayn et al. 2004; Nowinski et al.
2005). This has introduced the hypothesis that activation of either STN projection neurons and/or pallidothalamic fibers could be responsible for the therapeutic effects of
STN DBS.
The results of this study suggest that both STN neurons and GPi fibers are activated during clinically effective STN DBS. Recent histological tracing studies have shown that the LF is composed of axons from the entire GPi (Parent et al. 2001; Parent and Parent 2004), and not just its medial (non-motor) part as previously believed (Kuo et al. 1973; Kim et al. 1976). Consequently, stimulating LF through contacts located in the
120 dorsal STN would mechanistically be equivalent to direct GPi stimulation, at least when considering GPi output to thalamus. Since the GPi fibers of the LF are running in a tightly packed bundle they can be effectively stimulated with lower charge injection than with electrodes in the GPi where the neurons are widely dispersed. However, our simulations show that for monkey R7160, where the cathode was located in the ventral
STN, relatively few GPi fibers were activated, yet the stimulation was clinically effective. Overall our results suggest that large-scale activation of GPi fibers may not be necessary for therapeutic benefit, and a recent clinical study found that stimulating in the white matter above the STN was less effective than stimulation of the dorsolateral STN border (Herzog et al. 2004).
The large difference in model predicted GPi fiber activation in the two monkeys was supported by experimental single unit recordings performed during STN DBS
(Hashimoto et al., 2003). The proportion of GPi cells exhibiting short latency,
presumably antidromic activation, during STN DBS was similar to our theoretical
predictions in both monkeys (R7160 - 18% vs R370 - 82% in the model; R7160 - 8% vs
R370 - 41% in the experiments). The discrepancies observed between the model and
experimental results could be related to a number of factors. Our analysis of the
experimental data required that at least 15% of the DBS pulses must exhibit short latency
activation in the GPi cell to consider it antidromically activated. This somewhat arbitrary
threshold was necessary to perform our analysis for a number of reasons. First, no cell
will respond in exactly the same manner for all stimulus pulses because of stochastic
biological variability in the neural response and inherent problems with extracellular
microelectrode recordings during DBS (artifact subtraction, signal fluctuations, unit
121 isolation, spike sorting, etc). Second, GPi neurons fire at relatively high frequencies with a mix of antidromic and orthodromic action potentials during STN DBS (Hashimoto et al.
2003). Orthodromic activity in the GPi cells will collide with antidromic signals generated by stimulation of the LF. Therefore, antidromic excitation will not be recorded for every action potential evoked by the stimulation. In addition, our experimental sample of GPi cells was limited and approximately half of the GPi exit fibers would be expected to travel via the ansa lenticularis, a fiber tract rostral to the LF and outside the reach of the DBS voltage spread. Nonetheless, the experimental results support our model conclusions that therapeutic STN DBS generated significantly different activation of GPi fibers in the two monkeys.
The importance of activation of either STN projection neurons or GPi fibers of passage on the therapeutic outcome of STN DBS may be fundamentally dictated by the precise electrode location within the STN region. This concept is supported by our simulations where small changes (±0.25 mm) in electrode position could strongly affect
GPi fiber and STN neuron activation (Fig. 4-8). Clinical studies that found LF/ZI stimulation to be effective could have placed DBS electrodes in locations more suitable for maximal GPi fiber activation. However, post-operative electrode localization techniques used in human studies are not accurate to a sub-millimeter level making it difficult to determine electrode position with high certainty. Furthermore, it is possible that more than one ‘optimal’ contact and set of stimulation parameters exists, but due to time-consuming programming methods currently used it is not possible to definitively address every clinically beneficial setting. The results of this study lay the foundation for a prospective and coupled, theoretical and experimental, analysis of STN DBS where
122 electrode location and stimulation parameter selection can be systematically manipulated
to address the impact of STN neuron and/or GPi fiber activation on therapeutic outcome.
Such future developments in DBS research will enable a more complete understanding of the optimal electrode location and give better insight into the neural targets of STN DBS.
123 Chapter 5: Summary and future directions
5.1 Summary
The motivation for this project grew out of a need to better understand how
neurons respond to electrical stimulation in the brain, specifically to the deep brain
stimulation used in treatment of Parkinson’s disease. Since the serendipitous finding of
Benabid and colleagues that stimulation mimics lesion effects (Benabid et al., 1987),
DBS quickly entered the clinical arena where it was first used to treat medically- refractory movement disorders and is now being investigated for several other
neuropsychiatric disorders. The basic scientific research has not kept up with the rapid
growth of DBS into a main stream therapy and many basic questions regarding
mechanisms of DBS are still unanswered.
The primary goal of this project was to investigate therapeutic neuronal target of
subthalamic DBS (STN DBS). Even though the STN is the neuroanatomical target for
DBS electrode implantation, due to its small size and diverse neural environment (several
fiber tracts and nuclei surround the STN), it is not clear what is the true target of
therapeutic stimulation. The answer to this question is important for several reasons.
First, finding the effective stimulation parameters (active contact, amplitude, pulse width
and frequency) for each patient is a difficult and time consuming process of trial and
error. If the actual neural target(s) were known, the stimulation parameter search could be
optimized to the specific neural population(s) that surround the DBS electrode. Such
selective stimulation would likely improve the therapeutic outcome and, just as
importantly, decrease potential side effects. Second, DBS stimulation technology is
124 currently one-size-fits-all – neither the electrodes nor stimulation pulsing paradigms are
theoretically optimized for any of the specific nuclei targeted by DBS (STN, GPi,
thalamus, etc.). Knowledge of the desired neural target would likely prompt the
development of new technology geared towards more discerning stimulation paradigms.
Third, identifying the therapeutic target of STN DBS for Parkinson’s disease would improve understanding of the mechanisms of DBS and better explain the pathophysiology of the disease itself. DBS has been most thoroughly researched as a treatment for PD, and the hope is that the lessons we learn in PD will be applicable to other disorders that are also treatable with DBS.
The strategy for undertaking this project was to combine computational and experimental approaches. Computational techniques have steadily gained acceptance into the area of neuroscience research and are now recognized as a useful tool to test hypotheses and explain observed phenomena. Our goal was to bridge the gap between theoretical predictions and experimental outcomes by directly linking model result to animal behavior. This required the development of anatomically and biophysically accurate models customized to the experimental animals. By combining the expertise of two laboratories, one a leader in DBS computer modeling (McIntyre) and the other a leader in the neurophysiological investigation of DBS in parkinsonian monkeys (Vitek), this goal was feasible.
5.1.1 Stereotactic neurosurgical navigation software
The first step in constructing a comprehensive computational model of DBS was to build a stage. Our stage was a 3-dimensional anatomical reconstruction of the basal
125 ganglia nuclei from a monkey. We quickly realized that such a construct would be useful not only as a platform for our model, but also as a tool to plan and analyze subsequent
DBS implantations. This is how the idea for Cicerone was born. Image-guided neurosurgical navigation tools are common in clinical practice, but even there microelectrode mapping is accomplished using hand-drawn coordinates and generic brain atlases. Our goal was to build a system that would utilize available imaging technology and automate pre-operative planning, improve MER visualization and collection and provide DBS volume of activation predictions.
An important objective was to make the software truly user friendly and accessible to people outside the development team. This was accomplished by providing an attractive graphical user interface and extensive manual (Appendix A). Cicerone has been used for DBS implantation in three animals and there are several more currently in the preoperative planning stage. The software is also being used at several other laboratories outside our institution.
5.1.2 DBS voltage field
An integral part of a computational DBS model is the representation of voltage field in space and time. Voltage, or more precisely the second-order derivative of voltage, drives ions through the neuronal membrane which ultimately activates (or inhibits) the cell. Numerical techniques for calculating voltage generated in the tissue by a DBS electrode have been extensively utilized in the literature; however, these methods have not been directly validated. Therefore, we performed a study to measure voltages around a DBS electrode implanted in the thalamus of a monkey. We compared the in vivo and in
126 vitro results to volume conductor electric field models. We confirmed our hypothesis and found that appropriately parameterized models can accurately represent both spatial and
temporal characteristics of DBS induced voltages.
Two findings that stood out as particularly interesting are the voltage drop across
electrode interface and variability in the electrode impedance. Voltage field models
commonly assume ideal electrode behavior meaning that there is no voltage drop across the electrode-tissue interface. In both in vitro and in vivo preparations we found that the
recoded voltages were significantly smaller than the initial model predictions. The
models were adjusted so that they more accurately matched experimental observations.
For the in vitro model this meant adding a resistive component at the electrode-saline
interface. The in vivo model required inclusion of an encapsulation sheath and the resistance of this layer had to be increased more than five-fold to adequately fit the in vivo results.
Impedance of the DBS electrode had a great impact on the measured fields. Since clinical DBS devices use a voltage-controlled method of stimulation, it is expected that the amount of current injected into the tissue is proportional to the electrode impedance.
However, it was interesting to observe the variability in daily impedance measurements, and their dependence on the time since the electrode’s implantation and the electrode’s stimulation history. Since the animal used in this study was stimulated only acutely, it is possible that the impedance variations were more pronounced than it would be in the case of chronic stimulation. During chronic conditions it is likely that the electrode impedance would be continually on the low end of the spectrum.
127 The impedance finding has two important implications, in research and clinical
practice. Animals in DBS studies, even if they have implantable pulse generators, are
usually stimulated only acutely, during neurophysiological or behavioral experiments.
We now know that the electrode impedance will decrease significantly during the first hour or so after the stimulation starts. This means that the observed effects will also vary during this time because of the impedance and resulting voltage field changes. Similar effects may also directly impact human DBS patients. Once implanted, the DBS electrode is usually not turned on until a month after the surgery. When a patient returns to the hospital for DBS programming, the electrode impedance is probably rather high and as soon as the stimulation starts it will begin decreasing. DBS programming is a difficult process where small changes in patient’s symptoms must be appreciated in order to set the optimal parameters. If during that initial programming the stimulation effects change because of impedance variations, this could potentially interfere with accuracy of clinical evaluation and therefore with parameter optimization. This problem could be further compounded by the possibility of a continuous decrease in impedance after the patient has left the clinic. This could result in increased voltage fields and induce side effects that were not present during the clinical programming. Therefore, in both research and clinical applications it may be beneficial to apply stimulation for a period of time before any evaluations are undertaken.
5.1.3 Computational analysis of STN and LF activation during therapeutic DBS
The preceding sections described two integral components of the comprehensive
DBS model: anatomical model and voltage field model. Adding biophysical models of
128 individual STN neurons and pallidothalamic (GPi) fibers finalized the DBS model and allowed us to simulate the neural response to therapeutic stimulation. The anatomical
nuclei and electrode locations were reconstructed from two monkeys who had undergone
DBS implantation and behavioral testing. We could then correlate activation of the model
neurons to behavioral effects (therapeutic vs. non-therapeutic) observed in the animals.
The results indicate that activation of nearly half of the STN neurons is sufficient
for the behavioral manifestation of the therapeutic effects. We confirmed our hypothesis
that STN projection neurons are primarily activated during therapeutic STN DBS.
Significant activation of GPi fibers was observed in only one of the two animals so large-
scale activation of GPi fibers may not be necessary to achieve therapeutic effect. This is
contrary to several clinical studies which suggest GPi fiber involvement (Starr et al.,
2002; Lanotte et al., 2002; Saint-Cyr et al., 2002; Voges et al., 2002; Hamel et al., 2003;
Yelnik et al., 2003; Herzog et al., 2004; Zonenshayn et al., 2004; Nowinski et al., 2005;
Yokoyama et al., 2006; Godinho et al., 2006) so this issue warrants further investigation.
It is important to note two major limitations in the human studies: 1) anatomical location
of the active electrode contact was approximated using some combination of imaging, microelectrode recording data and brain atlases, but in most cases they did not include the
definitive histological verification; and 2) only the anatomical location of the active
electrode contact was used in the analysis. As a result, these studies were unable to
estimate the spread of stimulation since the stimulation parameters and electrode-tissue
properties were not considered. Our results indicate that sub-millimeter variations in the
electrode position in the STN region can strongly effect recruitment of surrounding
neural populations so precise determination of the electrode position is important.
129 Furthermore, we have shown that stimulus parameters and electrode-tissue electrical
properties will significantly affect the DBS voltage field and neural activation volume.
Therefore, it is imperative to consider these factors when predicting which neural populations are affected by stimulation.
Any results obtained from the model are useful only if the model can be properly
validated. This means that both individual model components and their interactions
accurately represent the biological system under investigation. Our intention was to use
the best possible “ingredients” available to us: 3D brain nuclei and electrode locations
were reconstructed directly from the histological brain slices; individual neuron model
geometries were based on histological tracing studies of STN neurons and GPi fibers; neuron model biophysical properties were derived from detailed electrophysiological studies of STN neurons and myelinated axons; and voltage field calculations were performed using advanced numerical techniques. However, the model was by no means perfect and certain limitations could not be avoided: histological slices were examined for only one of the two monkeys (electrode location for the second animal was derived from published anatomical reconstructions); only a small number of unique neuron geometries
was used and the number of modeled neurons (~100) was incomparable to the real
situation; STN neuron model biophysical characteristics were derived from experimental
recordings in rat neurons because similar data is not available for primates; axon models
were based on motor neuron fibers because myelinated axons of the basal ganglia
neurons have not been characterized in detail; and the voltage field models had not yet
been validated at the time of this study. Implications of these limitations are discussed in
detail in Chapter 4.
130 Despite the limitations, the comprehensive model was able to reproduce two key
experimental observations that were used to validate the model as a whole: corticospinal
tract activation thresholds and GPi neuron activation levels. CST fibers were included in
the model to serve as a neural population whose activation could be easily measured
experimentally. In the model we calculated stimulus amplitude necessary to activate a
substantial number of CST fibers and compared this value to the one obtained
experimentally. It was difficult to define what a ‘substantial number of fibers’ would be,
but we were satisfied that no CST fibers were activated at therapeutic amplitudes (when
no CST activation was observed experimentally) and ~10% of model fibers were active
at experimental CST thresholds. The difference seemed substantial enough for the gross
level validation that we were after with this method.
Previously published experimental study of DBS effects in these two monkeys included microelectrode recordings of GPi neurons during STN stimulation (Hashimoto
et al., 2003). This provided us with the second way of validating our model findings. We
reanalyzed the experimental data to determine what percentage of GPi neurons
experienced short-latency activation under different stimulation conditions. We assumed
that short-latency activation (< 1.5 ms from the stimulus pulse) would indicate antidromic
activation, meaning that the axon was activated first and the action potential traveled
antidromically to invade the cell body. Since STN neurons project to GPi we used this
method to differentiate between direct GPi neuron activation and synaptically-induced
activation due to STN neuron firing (this would presumably take longer than 1.5 ms).
We found that the model predictions of GPi fiber activation were in close agreement with
the experimental findings.
131 We have shown that anatomically and biophysically realistic models of DBS can reproduce certain experimental findings in an animal to which the model was customized.
We can therefore use such models to make reasonable quantitative predictions about events that cannot be observed directly with experimental techniques, and to guide further experimental studies. This study brings us one step closer to understanding the neural response to therapeutic stimulation, and paves way for utilizing such models for other DBS applications and electrical stimulation modalities.
5.2 Future directions
The use of DBS for treatment of nervous system disorders is expanding rapidly, and it is an exciting research area that will likely continue to grow. The use of computational models has proven to be effective in quantifying the stimulation effects, and the combination of custom computational models and experimental preparations has yielded valuable results. There are several interesting research problems that warrant further investigation.
The frequency of stimulation is a crucial parameter in DBS applications.
Frequencies above 100 Hz are generally considered to be therapeutic although the precise number varies across patients. One of the main drawbacks of our current DBS model is that frequency plays little role in the model results. The purpose of the model was to predict how far stimulation effects spread, in other words, what percentage of neurons are activated. Stimulation amplitude, pulse width and location of the active contact significantly affect the resulting volume of activation. The influence of stimulation
132 frequency however is minimal - if a neuron is close enough to the electrode to respond to low frequency stimulation it is equally likely to follow high frequency stimulation (high frequency is still slow enough so that neuron’s refractory period does not come into play). One possibility that the model does not consider is rapid accumulation of ions in the extracellular space that could be induced by high frequency discharge and inhibition of further firing. This explanation along with depolarization block and synaptic depletion
(Beurrier et al., 2001; Bikson et al., 2001; Urbano et al., 2002) have formed some of the early hypotheses on mechanisms of DBS. More recently it has become clear that high frequency firing does get transmitted to the downstream nuclei (Hashimoto et al., 2003;
Anderson et al., 2003; Maurice et al., 2003). Therefore, it is likely that the frequency plays a role in the stimulation-induced signals that are transmitted through the basal ganglia network (Grill et al., 2004). Since the current DBS modeling system does not consider connections between the neurons and propagation of signals between the nuclei, the effects of stimulation frequency cannot be appreciated. So the next step in the modeling process should involve creating a network of neurons downstream and upstream of the stimulation site. Some of these goals are already underway in the lab and they will greatly improve utility and relevance of the model results.
Voltage field measurements described in Chapter 3 were an important step in validating the field-neuron models used for analysis of DBS. But how the induced field affects neurons has not been directly addressed. The question of what is being stimulated will ultimately have to be confirmed through microelectrode recording studies.
Validation of GPi fiber activation described in Chapter 4 indicates how such task may be
133 accomplished. Recordings at multiple locations will reveal which neural populations are
being affected.
The induced voltage fields vary with the electrode impedance which in turn
experiences significant variations following the implantation. This confirms the idea that
voltage-controlled stimulation is not the optimal mode of stimulation for electrode
implanted in the tissue. Instead current-controlled stimulators would be more appropriate
as the current delivered would not depend on the electrode impedance, and the observed
effects are likely to be more predictable. Efforts are underway by several companies to bring such stimulators to the market. If and when a switch is made from voltage- controlled stimulation, it will be necessary to investigate effects of current-controlled
stimulation on the electrode and tissue properties.
One of the intriguing characteristic of DBS is the period of time necessary to
achieve full reduction of symptoms once stimulation is initiated and the mechanism underlying the prolonged therapeutic effect once stimulation is stopped (Temperli et al.,
2003). Recording experiments show that neural activity at the site of stimulation or in the site receiving projections from the stimulated site returns to baseline within milliseconds or seconds after stimulation stops. Yet, it may take minutes, hours or even days in some cases for symptoms to worsen. Similarly, when stimulation is initiated, improvement in gait may take hours to occur whereas tremor may disappear almost instantly. To account for this observation one would seemingly need to propose that there are changes occurring within the network over different timelines. This issue is even more relevant to other disorders treated with DBS, such as dystonia, depression and Tourette syndrome, where symptoms may take weeks to months to improve. To understand this process it
134 will be necessary to perform experiments over long time periods while recording from multiple neurons simultaneously at multiple sites. It is not yet clear what causes these effects although molecular studies have revealed that stimulation induces transcription of neurotransmitter-related genes and immediate early gene encoded proteins (Salin et al.,
2002; Bacci et al., 2004). Growth of new synapses could be a part of this process and this is a hypothesis that could be tested using a computational network model.
135 BIBLIOGRAPHY
Abosch A, Kapur S, Lang AE, Hussey D, Sime E, Miyasaki J, Houle S, Lozano AM. Stimulation of the subthalamic nucleus in Parkinson's disease does not produce striatal dopamine release. Neurosurgery. 2003;53(5):1095-102.
Alexander GE, Crutcher MD. Functional architecture of basal ganglia circuits: neural substrates of parallel processing. Trends Neurosci. 1990 Jul;13(7):266-71.
Alvarez-Royo P, Clower RP, Zola-Morgan S, Squire LR. Stereotaxic lesions of the hippocampus in monkeys: determination of surgical coordinates and analysis of lesions using magnetic resonance imaging. J Neurosci Methods., 1991; 38(2-3):223-32.
Anderson ME, Postupna N, Ruffo M. Effects of high-frequency stimulation in the internal globus pallidus on the activity of thalamic neurons in the awake monkey. J Neurophysiol 2003;89:1150–60.
Anderson TR, Hu B, Iremonger K, Kiss ZH. Selective attenuation of afferent synaptic transmission as a mechanism of thalamic deep brain stimulation-induced tremor arrest. J Neurosci. 26(3):841-50, 2006.
Asahi T, Tamura R, Eifuku S, Hayashi N, Endo S, Nishijo H, Ono T. A method for accurate determination of stereotaxic coordinates in single-unit recording studies in monkeys by high-resolution three-dimensional magnetic resonance imaging. Neurosci Res., 2003; 47(2):255-60.
Ashby P, Kim YJ, Kumar R, Lang AE, Lozano AM. Neurophysiological effects of stimulation through electrodes in the human subthalamic nucleus. Brain. 1999 Oct;122 ( Pt 10):1919-31.
Ashby P, Paradiso G, Saint-Cyr JA, Chen R, Lang AE, Lozano AM. Potentials recorded at the scalp by stimulation near the human subthalamic nucleus. Clin Neurophysiol. 2001; 112(3):431-7.
Ashby P, Strafella A, Dostrovsky JO, Lozano A, Lang AE. (1998) Immediate motor effects of stimulation through electrodes implanted in the human globus pallidus. Stereotact Funct Neurosurg. 70(1):1-18.
Astrom M, Johansson JD, Hariz MI, Eriksson O, Wardell K. The effect of cystic cavities on deep brain stimulation in the basal ganglia: a simulation-based study. J Neural Eng. 2006 Jun;3(2):132-8.
Aziz TZ, Peggs D, Sambrook MA, Crossman AR. Lesion of the subthalamic nucleus for the alleviation of 1-methyl-4-phenyl-1,2,3,6- tetrahydropyridine (MPTP)-induced parkinsonism in the primate. Mov Disord. 1991; 6(4):288-92.
Bacci JJ, Absi el H, Manrique C, Baunez C, Salin P, Kerkerian-Le Goff L. Differential effects of prolonged high frequency stimulation and of excitotoxic lesion of the subthalamic nucleus on dopamine denervation-induced cellular defects in the rat striatum and globus pallidus. Eur J Neurosci. 2004;20(12):3331-41.
Baker KB, Montgomery EB Jr, Rezai AR, Burgess R, Luders HO. Subthalamic nucleus deep brain stimulus evoked potentials: physiological and therapeutic implications. Mov Disord. 2002 Sep;17(5):969-83.
Baldissera F, Lundberg A, Udo M. Stimulation of pre- and postsynaptic elements in the red nucleus. Exp Brain Res 1972;15:151–67.
Bar-Gad I, Elias S, Vaadia E, Bergman H. Complex locking rather than complete cessation of neuronal activity in the globus pallidus of a 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-treated primate in response to pallidal microstimulation. J Neurosci. 2004;24(33):7410-9.
Bejjani B, Damier P, Arnulf I, Bonnet AM, Vidailhet M, Dormont D, Pidoux B, Cornu P, Marsault C, Agid Y. Pallidal stimulation for Parkinson's disease. Two targets? Neurology. 1997; 49(6):1564-9.
Benabid AL, Pollak P, Gao D, Hoffmann D, Limousin P, Gay E, Payen I, Benazzouz A. (1996) Chronic electrical stimulation of the ventralis intermedius nucleus of the thalamus as a treatment of movement disorders. J. Neurosurg. 84 203–14.
Benabid AL, Pollak P, Gervason C, Hoffmann D, Gao DM, Hommel M, Perret JE, de Rougemont J. Long-term suppression of tremor by chronic stimulation of the ventral intermediate thalamic nucleus. Lancet. 1991; 337(8738):403-6.
Benabid AL, Pollak P, Louveau A, Henry S, de Rougemont J. Combined (thalamotomy and stimulation) stereotactic surgery of the VIM thalamic nucleus for bilateral Parkinson disease. Appl Neurophysiol 1987;50:344–6.
Benabid AL. (2003) Deep brain stimulation of Parkinson’s disease. Curr Opin Neurobiol 13(6):696-706.
Benazzouz A, Breit S, Koudsie A, Pollak P, Krack P, Benabid AL. Intraoperative microrecordings of the subthalamic nucleus in Parkinson's disease. Mov Disord. 2002; 17 Suppl 3:S145-9.
136 Benazzouz A, Gao DM, Ni ZG, Piallat B, Bouali-Benazzouz R, Benabid AL. Effect of high-frequency stimulation of the subthalamic nucleus on the neuronal activities of the substantia nigra pars reticulata and ventrolateral nucleus of the thalamus in the rat. Neuroscience. 2000; 99(2):289-95.
Benazzouz A, Gross C, Feger J, Boraud T, Bioulac B. Reversal of rigidity and improvement in motor performance by subthalamic high-frequency stimulation in MPTP-treated monkeys. Eur J Neurosci. 1993; 5(4):382-9.
Benazzouz A, Piallat B, Pollak P, Benabid AL. Responses of substantia nigra pars reticulata and globus pallidus complex to high frequency stimulation of the subthalamic nucleus in rats: electrophysiological data. Neurosci Lett. 1995; 189(2):77-80.
Benazzouz A, Tai CH, Meissner W, Bioulac B, Bezard E, Gross C. High-frequency stimulation of both zona incerta and subthalamic nucleus induces a similar normalization of basal ganglia metabolic activity in experimental parkinsonism. FASEB J. 2004 ; 18(3):528- 30.
Bergman H, Feingold A, Nini A, Raz A, Slovin H, Abeles M, Vaadia E. (1998) Physiological aspects of information processing in the basal ganglia of normal and parkinsonian primates. Trends Neurosci. 21:32-38.
Bergman H, Wichmann T, DeLong MR. Reversal of experimental parkinsonism by lesions of the subthalamic nucleus. Science. 1990; 249(4975):1436-8.
Bergman H, Wichmann T, Karmon B, DeLong MR. The primate subthalamic nucleus. II. Neuronal activity in the MPTP model of parkinsonism. J Neurophysiol. 1994; 72(2):507-20.
Beurrier C, Bioulac B, Audin J, Hammond C. High-frequency stimulation produces a transient blockade of voltage-gated currents in subthalamic neurons. J Neurophysiol. 2001; 85(4):1351-6.
Beurrier C, Bioulac B, Hammond C. (2000) Slowly inactivating sodium current (I(NaP)) underlies single-spike activity in rat subthalamic neurons. J Neurophysiol. 83:1951-1957.
Beurrier C, Congar P, Bioulac B, Hammond C. Subthalamic nucleus neurons switch from single-spike activity to burst-firing mode. J Neurosci. 19(2):599-609, 1999.
Bevan MD, Bolam JP. (1995) Cholinergic, GABAergic, and glutamate-enriched inputs from the mesopontine tegmentum to the subthalamic nucleus in the rat. J Neurosci. 15:7105-120.
Bevan MD, Magill PJ, Hallworth NE, Bolam JP, Wilson CJ. (2002) Regulation of the timing and pattern of action potential generation in rat subthalamic neurons in vitro by GABA-A IPSPs. J Neurophysiol. 87:1348-1362.
Bevan MD, Wilson CJ. Mechanisms underlying spontaneous oscillation and rhythmic firing in rat subthalamic neurons. J Neurosci. 1999;19(17):7617-28.
Bikson M, Lian J, Hahn PJ, Stacey WC, Sciortino C, Durand DM. (2001) Suppression of epileptiform activity by high frequency sinusoidal fields in rat hippocampal slices. J Physiol. 531:181-191.
Bookstein FL. Morphometrics. In Toga AW editor. Three-dimensional Neuroimaging, Ch 8. Raven Press, New York, 1990.
Boraud T, Bezard E, Bioulac B, Gross C. High-frequency stimulation of the internal Globus Pallidus (GPi) simultaneously improves parkinsonian symptoms and reduces the firing frequency of GPi neurons in the MPTP-treated monkey. Neurosci Lett 1996;215:17–20.
Boulet S, Lacombe E, Carcenac C, Feuerstein C, Sgambato-Faure V, Poupard A, Savasta M. Subthalamic stimulation-induced forelimb dyskinesias are linked to an increase in glutamate levels in the substantia nigra pars reticulata. J Neurosci. 2006;26(42):10768-76.
Brown P, Mazzone P, Oliviero A, Altibrandi MG, Pilato F, Tonali PA, Di Lazzaro V. Effects of stimulation of the subthalamic area on oscillatory pallidal activity in Parkinson's disease. Exp Neurol. 2004;188(2):480-90.
Brown P, Oliviero A, Mazzone P, Insola A, Tonali P, Di Lazzaro V. Dopamine dependency of oscillations between subthalamic nucleus and pallidum in Parkinson's disease. J Neurosci. 2001;21(3):1033-8.
Bruet N, Windels F, Bertrand A, Feuerstein C, Poupard A, Savasta M. High-frequency stimulation of the subthalamic nucleus increases the extracellular contents of striatal dopamine in normal and partially dopaminergic denervated rats. J Neuropathol Exp Neurol 2001;60: 15–24.
Burchiel KJ, Anderson VC, Favre J, Hammerstad JP. Comparison of pallidal and subthalamic nucleus deep brain stimulation for advanced Parkinson's disease: results of a randomized, blinded pilot study. Neurosurgery 45(6):1375-82, 1999.
Butson CR, Cooper SE, Henderson JM, McIntyre CC. Patient-specific analysis of the volume of tissue activated during deep brain stimulation. Neuroimage. 2007;34(2):661-70.
137 Butson CR, Maks CB, McIntyre CC. Sources and effects of electrode impedance during deep brain stimulation. Clin Neurophysiol., 2006a;117(2):447-54.
Butson CR, McIntyre CC. Role of electrode design on the volume of tissue activated during deep brain stimulation. J Neural Eng., 2006; 3(1):1-8.
Butson CR, McIntyre CC. Tissue and electrode capacitance reduce neural activation volumes during deep brain stimulation. Clin Neurophysiol., 2005; 116(10):2490-500.
Canavero S, Paolotti R, Bonicalzi V, Castellano G, Greco-Crasto S, Rizzo L, Davini O, Zenga F, Ragazzi P. Extradural motor cortex stimulation for advanced Parkinson disease. Report of two cases. J Neurosurg. 2002; 97(5):1208-11.
Caparros-Lefebvre D, Ruchoux MM, Blond S, Petit H, Percheron G. Long-term thalamic stimulation in Parkinson's disease: postmortem anatomoclinical study. Neurology.1994 Oct;44(10):1856-60.
Carpenter MB, Carleton SC, Keller JT, Conte P. (1981) Connections of the subthalamic nucleus in the monkey. Brain Res. 224(1):1- 29.
Ceballos-Baumann AO. Functional imaging in Parkinson's disease: activation studies with PET, fMRI and SPECT. J Neurol. 2003;250 Suppl 1:I15-23.
Choi AQ, Cavanaugh JK, Durand DM. Selectivity of multiple-contact nerve cuff electrodes: a simulation analysis. IEEE Trans Biomed Eng. 2001 Feb;48(2):165-72.
Christensen GE, Joshi SC, Miller MI. Volumetric transformation of brain anatomy. IEEE Trans Med Imaging., 1997; 16(6):864-77.
Coburn B, Sin WK. A theoretical study of epidural electrical stimulation of the spinal cord--Part I: Finite element analysis of stimulus fields. IEEE Trans Biomed Eng. 1985 Nov;32(11):971-7.
DeLong MR, Crutcher MD, Georgopoulos AP. Primate globus pallidus and subthalamic nucleus: functional organization. J Neurophysiol., 1985; 53(2):530-43.
DeLong MR. Activity of pallidal neurons during movement. J Neurophysiol., 1971; 34(3):414-27.
DeLong MR. Primate models of movement disorders of basal ganglia origin. Trends Neurosci. 1990;13(7):281-5
Deogaonkar M, Heers M, Mahajan S, Brummer M, Subramanian T. Method of construction of a MRI-based tabular database of 3D stereotaxic co-ordinates for individual structures in the basal ganglia of Macaca mulatta. J Neurosci Methods., 2005; 149(2):154-63.
Destexhe A, Mainen ZF, and Sejnowski TJ. An efficient method for computing synaptic conductances based on a kinetic model of receptor binding. Neural Comput 6: 14–18, 1994a.
Destexhe A, Mainen ZF, and Sejnowski TJ. Synthesis of models for excitable membranes, synaptic transmission and neuromodulation using a common kinetic formalism. J Comput Neurosci 1: 195–230, 1994b.
Deuschl G, Raethjen J, Lindemann M, Krack P. (2001) The pathophysiology of tremor. Muscle Nerve. 24:716-735.
D'Haese PF, Cetinkaya E, Konrad PE, Kao C, Dawant BM. Computer-aided placement of deep brain stimulators: from planning to intraoperative guidance. IEEE Trans Med Imaging., 2005; 24(11):1469-78.
Do MT, Bean BP. Subthreshold sodium currents and pacemaking of subthalamic neurons: modulation by slow inactivation. Neuron. 39(1):109-20, 2003.
Dostrovsky JO, Levy R, Wu JP, Hutchison WD, Tasker RR, Lozano AM. Microstimulation-induced inhibition of neuronal firing in human globus pallidus. J Neurophysiol 2000;84:570–4.
Dostrovsky JO, Lozano AM. Mechanisms of deep brain stimulation. Mov Disord. 2002;17 Suppl 3:S63-8.
Elder CM, Hashimoto T, Zhang J, Vitek JL. Chronic implantation of deep brain stimulation leads in animal models of neurological disorders. J Neurosci Methods., 2005; 142(1):11-6.
Filali M, Hutchison WD, Palter VN, Lozano AM, Dostrovsky JO. Stimulation-induced inhibition of neuronal firing in human subthalamic nucleus. Exp Brain Res. 2004;156(3):274-81.
Finnis KW, Starreveld YP, Parrent AG, Sadikot AF, Peters TM. Three-dimensional database of subcortical electrophysiology for image-guided stereotactic functional neurosurgery. IEEE Trans Med Imaging., 2003; 22(1):93-104.
Foffani G, Ardolino G, Egidi M, Caputo E, Bossi B, Priori A. Subthalamic oscillatory activities at beta or higher frequency do not change after high-frequency DBS in Parkinson's disease. Brain Res Bull. 2006;69(2):123-30.
138 Foffani G, Priori A, Egidi M, Rampini P, Tamma F, Caputo E, Moxon KA, Cerutti S, Barbieri S. 300-Hz subthalamic oscillations in Parkinson's disease. Brain. 2003;126(Pt 10):2153-63.
Ford B, Winfield L, Pullman SL, Frucht SJ, Du Y, Greene P, Cheringal JH, Yu Q, Cote LJ, Fahn S, McKhann GM 2nd, Goodman RR. (2004) Subthalamic nucleus stimulation in advanced Parkinson's disease: blinded assessments at one year follow up. J Neurol Neurosurg Psychiatry. 75(9):1255-9.
Francois C, Yelnik J, Percheron G. A stereotaxic atlas of the basal ganglia in macaques. Brain Res Bull., 1996; 41(3):151-8.
Frey S, Comeau R, Hynes B, Mackey S, Petrides M. Frameless stereotaxy in the nonhuman primate. Neuroimage. 2004; 23(3):1226- 34.
Gabriels L, Cosyns P, Nuttin B, Demeulemeester H, Gybels J. (2003) Deep brain stimulation for treatment-refractory obsessive- compulsive disorder: psychopathological and neuropsychological outcome in three cases. Acta Psychiatr. Scand. 107 275–82.
Galati S, Mazzone P, Fedele E, Pisani A, Peppe A, Pierantozzi M, Brusa L, Tropepi D, Moschella V, Raiteri M, Stanzione P, Bernardi G, Stefani A. Biochemical and electrophysiological changes of substantia nigra pars reticulata driven by subthalamic stimulation in patients with Parkinson's disease. Eur J Neurosci. 2006;23(11):2923-8.
Garcia L, Audin J, D’Alessandro G, Bioulac B, Hammond C. Dual effect of high-frequency stimulation on subthalamic neuron activity. J Neurosci 2003;23:8743–51.
Garcia L, D'Alessandro G, Fernagut PO, Bioulac B, Hammond C. Impact of high-frequency stimulation parameters on the pattern of discharge of subthalamic neurons. J Neurophysiol. 2005;94(6):3662-9.
Gillies A, Willshaw D. Membrane channel interactions underlying rat subthalamic projection neuron rhythmic and bursting activity. J Neurophysiol. 95(4):2352-65, 2006.
Gimsa U, Schreiber U, Habel B, Flehr J, van Rienen U, Gimsa J. Matching geometry and stimulation parameters of electrodes for deep brain stimulation experiments--numerical considerations. J Neurosci Methods. 2006 Jan 30;150(2):212-27.
Gironell A, Amirian G, Kulisevsky J, Molet J. (2005) Usefulness of an intraoperative electrophysiological navigator system for subthalamic nucleus surgery in Parkinson's disease. Stereotact Funct Neurosurg. 83(2-3):101-7.
Godinho F, Thobois S, Magnin M, Guenot M, Polo G, Benatru I, Xie J, Salvetti A, Garcia-Larrea L, Broussolle E, Mertens P. Subthalamic nucleus stimulation in Parkinson's disease : anatomical and electrophysiological localization of active contacts. J Neurol. 2006;253(10):1347-55.
Grandori F, Rossini P. Electrical stimulation of the motor cortex: theoretical considerations. Ann Biomed Eng. 1988;16(6):639-52.
Greenberg BD, Rezai AR. (2003) Mechanisms and the current state of deep brain stimulation in neuropsychiatry. CNS Spectr 8(7):522-6.
Grill WM, McIntyre CC. (2001) Extracellular excitation of central neurons: implications for the mechanisms of deep brain stimulation. Thalamus Rel Syst. 1:269-277.
Grill WM, Mortimer JT. Electrical properties of implant encapsulation tissue. Ann Biomed Eng. 22(1):23-33, 1994.
Grill WM, Snyder AN, Miocinovic S. Deep brain stimulation creates an informational lesion of the stimulated nucleus. Neuroreport. 2004;15(7):1137-40.
Gustafsson B, Jankowska E. Direct and indirect activation of nerve cells by electrical pulses applied extracellularly. J Physiol. 1976 Jun;258(1):33-61, 1976.
Haberler C, Alesch F, Mazal PR, Pilz P, Jellinger K, Pinter MM, Hainfellner JA, Budka H. No tissue damage by chronic deep brain stimulation in Parkinson's disease. Ann Neurol. 2000 Sep;48(3):372-6.
Hanekom T. Modelling encapsulation tissue around cochlear implant electrodes. Med Biol Eng Comput. 2005 Jan;43(1):47-55.
Hamani C, Saint-Cyr JA, Fraser J, Kaplitt M, Lozano AM. The subthalamic nucleus in the context of movement disorders. Brain. 2004;127(Pt 1):4-20.
Hamel W, Fietzek U, Morsnowski A, Schrader B, Herzog J, Weinert D, Pfister G, Muller D, Volkmann J, Deuschl G, Mehdorn HM. Deep brain stimulation of the subthalamic nucleus in Parkinson's disease: evaluation of active electrode contacts. J Neurol Neurosurg Psychiatry. 2003;74(8):1036-46.
Hardman CD, Henderson JM, Finkelstein DI, Horne MK, Paxinos G, Halliday GM. (2002) Comparison of the basal ganglia in rats, marmosets, macaques, baboons, and humans: volume and neuronal number for the output, internal relay, and striatal modulating nuclei. J Comp Neurol. 445(3):238-55.
139 Hardy P, Barnett G. Spatial distortion in magnetic resonance imaging: impact on stereotactic localization. In Gildenberg PL, Tasker RR, editors. Textbook of Stereotactic and Functional Neurosurgery, Ch 33. New York: McGraw-Hill, 1998.
Hashimoto T, Elder CM, Okun MS, Patrick SK, Vitek JL. Stimulation of the subthalamic nucleus changes the firing pattern of pallidal neurons. J Neurosci., 2003; 23(5):1916-23.
Hashimoto T, Elder CM, Vitek JL. A template subtraction method for stimulus artifact removal in high-frequency deep brain stimulation. J Neurosci Methods 2002;113:181–6.
Hastreiter P, Rezk-Salama C, Soza G, Bauer M, Greiner G, Fahlbusch R, Ganslandt O, Nimsky C. Strategies for brain shift evaluation. Med Image Anal., 2004; 8(4):447-64.
Haueisen J, Tuch DS, Ramon C, Schimpf PH, Wedeen VJ, George JS, Belliveau JW. (2002) The influence of brain tissue anisotropy on human EEG and MEG. Neuroimage. 15:159-166.
Hemm S, Mennessier G, Vayssiere N, Cif L, El Fertit H, Coubes P. Deep brain stimulation in movement disorders: stereotactic coregistration of two-dimensional electrical field modeling and magnetic resonance imaging. J Neurosurg. 2005 Dec;103(6):949-55.
Hemm S, Vayssiere N, Mennessier G, Cif L, Zanca M, Ravel M, Frerebeau P, Coubes P, Evolution of brain impedance in dystonic patients treated by gpi electrical stimulation, Neuromodulation 7 (2) (2004) 67–75.
Hershey T, Revilla FJ, Wernle AR, McGee-Minnich L, Antenor JV, Videen TO, Dowling JL, Mink JW, Perlmutter JS. Cortical and subcortical blood flow effects of subthalamic nucleus stimulation in PD. Neurology. 2003;61(6):816-21.
Herzog J, Fietzek U, Hamel W, Morsnowski A, Steigerwald F, Schrader B, Weinert D, Pfister G, Muller D, Mehdorn HM, Deuschl G, Volkmann J. Most effective stimulation site in subthalamic deep brain stimulation for Parkinson's disease. Mov Disord. 2004;19(9):1050-4.
Hilker R, Voges J, Ghaemi M, Lehrke R, Rudolf J, Koulousakis A, Herholz K, Wienhard K, Strum V, Heiss WD. Deep brain stimulation of the subthalamic nucleus does not increase the striatal dopamine concentration in parkinsonian humans. Mov Disord 2003;18:41–8.
Hines ML, Carnevale NT. The NEURON simulation environment. Neural Comput., 1997; 9(6):1179-209.
Hodaie M, Wennberg RA, Dostrovsky JO, Lozano AM. (2002) Chronic anterior thalamus stimulation for intractable epilepsy. Epilepsia 43 603–8.
Hoekema R, Venner K, Struijk JJ, Holsheimer J. Multigrid solution of the potential field in modeling electrical nerve stimulation. Comput Biomed Res. 1998 Oct;31(5):348-62.
Holsheimer J. Computer modelling of spinal cord stimulation and its contribution to therapeutic efficacy. Spinal Cord. 1998 Aug;36(8):531-40.
Holsheimer J, Demeulemeester H, Nuttin B, de Sutter P. (2000) Identification of the target neuronal elements in electrical deep brain stimulation. European J. Neuroscience 12:4573-4577.
Hutchison WD, Allan RJ, Opitz H, Levy R, Dostrovsky JO, Lang AE, Lozano AM. Neurophysiological identification of the subthalamic nucleus in surgery for Parkinson's disease. Ann Neurol. 1998; 44(4):622-8.
Hutchison WD, Chung AG, Goldshmidt A. (2002) Chronaxie and refractory period of neuronal inhibition by extracellular stimulation in the region of rat STN. Program No. 416.3. Abstract Viewer/Itinerary Planner CD-ROM, Society for Neuroscience, Washington, DC.
Ilinsky IA, Kultas-Ilinsky K, Knosp B. Stereotactic Atlas of the Macaca Mulatta Thalamus and Adjacent Basal Ganglia Nuclei. Springer, New York, 2002.
Jankowska E, Padel Y, Tanaka R. The mode of activation of pyramidal tract cells by intracortical stimuli. J Physiol 1975;249:617–36.
Jech R, Urgosik D, Tintera J, Nebuzelsky A, Krasensky J, Liscak R, Roth J, Ruzicka E. Functional magnetic resonance imaging during deep brain stimulation: a pilot study in 4 patients with Parkinson’s disease. Mov Disord 2001;16:1126–32.
Johnson MD, Otto KJ, Kipke DR. Repeated voltage biasing improves unit recordings by reducing resistive tissue impedances. IEEE Trans Neural Syst Rehabil Eng. 2005 Jun;13(2):160-5.
Kim R, Nakano K, Jayaraman A, Carpenter MB. Projections of the globus pallidus and adjacent structures: an autoradiographic study in the monkey. J Comp Neurol. 169(3):263-90, 1976.
140 Kiss ZHT, Mooney DM, Renaud L, Hu B. (2002) Neuronal response to local electrical stimulation in rat thalamus: physiological implications for mechanisms of deep brain stimulation. Neuroscience. 113:137-143.
Kita H, Chang HT, Kitai ST. Pallidal inputs to subthalamus: intracellular analysis. Brain Res. 264(2):255-65, 1983.
Kita H, Tachibana Y, Nambu A, Chiken S. Balance of monosynaptic excitatory and disynaptic inhibitory responses of the globus pallidus induced after stimulation of the subthalamic nucleus in the monkey. J Neurosci. 2005;25(38):8611-9.
Kleiner-Fisman G, Herzog J, Fisman DN, Tamma F, Lyons KE, Pahwa R, Lang AE, Deuschl G. Subthalamic nucleus deep brain stimulation: summary and meta-analysis of outcomes. Mov Disord. 2006; 21 Suppl 14:S290-304.
Krack P, Batir A, Van Blercom N, Chabardes S, Fraix V, Ardouin C, Koudsie A, Limousin PD, Benazzouz A, LeBas JF, Benabid AL,
Pollak P. Five-year follow-up of bilateral stimulation of the subthalamic nucleus in advanced Parkinson's disease. N Engl J Med. 349(20):1925-34, 2003.
Krack P, Fraix V, Mendes A, Benabid AL, Pollak P. (2002) Postoperative management of subthalamic nucleus stimulation for Parkinson's disease. Mov Disord. 17 Suppl 3:S188-97.
Kumar R, Lozano AM, Montgomery E, Lang AE. Pallidotomy and deep brain stimulation of the pallidum and subthalamic nucleus in advanced Parkinson's disease. Mov Disord. 1998; 13 Suppl 1:73-82.
Kuo JS, Carpenter MB. Organization of pallidothalamic projections in the rhesus monkey. J Comp Neurol. 151(3):201-36, 1973.
Lanotte MM, Rizzone M, Bergamasco B, Faccani G, Melcarne A, Lopiano L. Deep brain stimulation of the subthalamic nucleus: anatomical, neurophysiological, and outcome correlations with the effects of stimulation. J Neurol Neurosurg Psychiatry. 2002;72(1):53-8.
Latimer DC, Roth BJ. Electrical stimulation of cardiac tissue by a bipolar electrode in a conductive bath. IEEE Trans Biomed Eng. 1998 Dec;45(12):1449-58.
Le Bihan D, Mangin JF, Poupon C, Clark CA, Pappata S, Molko N, Chabriat H. (2001) Diffusion tensor imaging: concepts and applications. J Magn Reson Imaging. 13:534-546.
Lee DC, McIntyre CC, and WM Grill. (2003) Extracellular electrical stimulation of central neurons: quantitative studies. In: Handbook of Neuroprosthetic Reseach Methods. WE Finn and PG Lopresti (eds) CRC Press, pp. 95-125.
Lee KH, Blaha CD, Harris BT, Cooper S, Hitti FL, Leiter JC, Roberts DW, Kim U. Dopamine efflux in the rat striatum evoked by electrical stimulation of the subthalamic nucleus: potential mechanism of action in Parkinson's disease. Eur J Neurosci. 2006 ;23(4):1005-14.
Lee KH, Chang SY, Roberts DW, Kim U. Neurotransmitter release from high-frequency stimulation of the subthalamic nucleus. J Neurosurg. 2004;101(3):511-7.
Lee KH, Roberts DW, Kim U. Effect of high-frequency stimulation of the subthalamic nucleus on subthalamic neurons: an intracellular study. Stereotact Funct Neurosurg. 2003;80(1-4):32-6.
Lee, DC, and Grill WM. (2001) Polarization of spherical cells by non-uniform electric fields: steady-state analysis. Ann Biomed Eng. 29:S128.
Levy R, Ashby P, Hutchison WD, Lang AE, Lozano AM, Dostrovsky JO. Dependence of subthalamic nucleus oscillations on movement and dopamine in Parkinson's disease. Brain. 2002;125(Pt 6):1196-209.
Lewis SJG, Caldwell MA, Barker RA. (2003) Modern therapeutic approaches in Parkinson’s disease. Expert Rev Mol Med 5:1-20.
Lian J, Bikson M, Sciortino C, Stacey WC, Durand DM. (2003) Local suppression of epileptiform activity by electrical stimulation in rat hippocampus in vitro. J Physiol 547:427-434.
Limousin P, Krack P, Pollak P, Benazzouz A, Ardouin C, Hoffmann D, Benabid AL. Electrical stimulation of the subthalamic nucleus in advanced Parkinson's disease. N Engl J Med. 339(16):1105-11, 1998.
Limousin P, Pollak P, Benazzouz A, Hoffmann D, Le Bas JF, Broussolle E, Perret JE, Benabid AL. Effect of parkinsonian signs and symptoms of bilateral subthalamic nucleus stimulation. Lancet. 1995; 345(8942):91-5.
Lozano AM, Mahant N. (2004) Deep brain stimulation surgery for Parkinson's disease: mechanisms and consequences. Parkinsonism Relat Disord 1:S49-57.
Maciunas RJ, Galloway RL Jr, Latimer JW. The application accuracy of stereotactic frames. Neurosurgery., 1994; 35(4):682-94.
141 Magarinos-Ascone C, Pazo JH, Macadar O, Buno W. High-frequency stimulation of the subthalamic nucleus silences subthalamic neurons: a possible cellular mechanism in Parkinson’s disease. Neuroscience 2002;115:1109–17.
Magill PJ, Bolam JP, Bevan MD. Dopamine regulates the impact of the cerebral cortex on the subthalamic nucleus-globus pallidus network. Neuroscience. 2001;106(2):313-30.
Magnin M, Morel A, Jeanmonod D. Single-unit analysis of the pallidum, thalamus and subthalamic nucleus in parkinsonian patients. Neuroscience. 2000;96(3):549-64.
Manola L, Roelofsen BH, Holsheimer J, Marani E, Geelen J. Modelling motor cortex stimulation for chronic pain control: electrical potential field, activating functions and responses of simple nerve fibre models. Med Biol Eng Comput. 2005 May;43(3):335-43.
Marsden JF, Limousin-Dowsey P, Ashby P, Pollak P, Brown P. Subthalamic nucleus, sensorimotor cortex and muscle interrelationships in Parkinson's disease. Brain. 2001; 124(Pt 2):378-88.
Martin RF, Bowden DM. Primate Brain Maps: Structure of the Macaque Brain. Elsevier: Amsterdam, 2000.
Maurice N, Thierry AM, Glowinski J, Deniau JM. Spontaneous and evoked activity of substantia nigra pars reticulata neurons during high-frequency stimulation of the subthalamic nucleus. J Neurosci 2003;23:9929–36.
Mayberg HS, Lozano AM, Voon V, McNeely HE, Seminowicz D, Hamani C, Schwalb JM, Kennedy SH. (2005) Deep brain stimulation for treatment-resistant depression. Neuron 45 651–60.
Mazzone P, Lozano A, Stanzione P, Galati S, Scarnati E, Peppe A, Stefani A. Implantation of human pedunculopontine nucleus: a safe and clinically relevant target in Parkinson's disease. Neuroreport. 2005; 16(17):1877-81.
McIntyre CC, Grill WM, Sherman DL, Thakor NV. Cellular effects of deep brain stimulation: model-based analysis of activation and inhibition. J Neurophysiol. 91(4):1457-69, 2004a.
McIntyre CC, Grill WM. (2000) Selective microstimulation of central nervous system neurons. Ann Biomed Eng. 28:219-233.
McIntyre CC, Grill WM. (2002) Extracellular stimulation of central neurons: influence of stimulus waveform and frequency on neuronal output. J Neurophysiol. 88:1592-1604.
McIntyre CC, Grill WM. Excitation of central nervous system neurons by nonuniform electric fields. Biophys J. 1999;76(2):878-88.
McIntyre CC, Mori S, Sherman DL, Thakor NV, Vitek JL. Electric field and stimulating influence generated by deep brain stimulation of the subthalamic nucleus. Clin Neurophysiol. 115(3):589-95, 2004b.
McIntyre CC, Richardson AG, Grill WM. Modeling the excitability of mammalian nerve fibers: influence of afterpotentials on the recovery cycle. J Neurophysiol. 87(2):995-1006, 2002.
McIntyre CC, Savasta M, Kerkerian-Le Goff L, Vitek JL. (2004d) Uncovering the mechanism(s) of action of deep brain stimulation: activation, inhibition, or both. Clin Neurophysiol 115:1239-1248.
McIntyre CC, Savasta M, Walter BL, Vitek JL. (2004c) How does deep brain stimulation work? Present understanding and future questions. J Clin Neurophysiol 21:40-50.
McNeal DR. Analysis of a model for excitation of myelinated nerve. IEEE Trans Biomed Eng. 1976; 23:329-337.
Meissner W, Harnack D, Hoessle N, Bezard E, Winter C, Morgenstern R, Kupsch A. High frequency stimulation of the entopeduncular nucleus has no effect on striatal dopaminergic transmission. Neurochem Int. 2004;44(4):281-6.
Meissner W, Harnack D, Paul G,Reum T, Sohr R, Morgenstern R, Kupsch A. Deep brain stimulation of subthalamic neurons increases striatal dopamine metabolism and induces contralateral circling in freely moving 6-hydroxydopamine-lesioned rats. Neurosci Lett 2002;328:105–8.
Meissner W, Harnack D, Reese R, Paul G, Reum T, Ansorge M, Kusserow H, Winter C, Morgenstern R, Kupsch A. High-frequency stimulation of the subthalamic nucleus enhances striatal dopamine release and metabolism in rats. J Neurochem 2003;85:601–9.
Meissner W, Leblois A, Hansel D, Bioulac B, Gross CE, Benazzouz A, Boraud T. Subthalamic high frequency stimulation resets subthalamic firing and reduces abnormal oscillations. Brain. 2005; 128(Pt 10):2372-82.
Miocinovic S, Grill WM. Sensitivity of temporal excitation properties to the neuronal element activated by extracellular stimulation. J Neurosci Methods. 2004 132(1):91-9.
Miocinovic S, Parent M, Butson CR, Hahn PJ, Russo GS, Vitek JL, McIntyre CC. Computational analysis of subthalamic nucleus and lenticular fasciculus activation during therapeutic deep brain stimulation. J Neurophysiol. 2006; 96(3):1569-80.
142 Miocinovic S, Zhang J, Xu W, Russo GS, Vitek JL, McIntyre CC. Stereotactic neurosurgical planning, recording, and visualization for deep brain stimulation in non-human primates. J Neurosci Methods. 2007 May 15;162(1-2):32-41.
Montgomery EB Jr. Effects of GPi stimulation on human thalamic neuronal activity. Clin Neurophysiol. 2006;117(12):2691-702.
Montgomery EB, Baker KB. Mechanisms of deep brain stimulation and future technical developments. Neurol Res 2000;22:259–66.
Moriizumi T, Nakamura Y, Kitao Y, Kudo M. (1987) Ultrastructural analyses of afferent terminals in the subthalamic nucleus of the cat with a combined degeneration and horseradish peroxidase tracing method. J Comp Neurol. 265:159-174.
Moss J, Ryder T, Aziz TZ, Graeber MB, Bain PG. Electron microscopy of tissue adherent to explanted electrodes in dystonia and Parkinson's disease. Brain. 2004 Dec;127(Pt 12):2755-63.
Nahm FK, Dale AM, Albright TD, Amaral DG. In vivo microelectrode localization in the brain of the alert monkey: a combined radiographic and magnetic resonance imaging approach. Exp Brain Res., 1994; 98(3):401-11.
Nakanishi H, Kita H, Kitai ST. Electrical membrane properties of rat subthalamic neurons in an in vitro slice preparation. Brain Res. 437(1):35-44, 1987.
Nambu A, Takada M, Inase M, Tokuno H. (1996) Dual somatotopical representations in the primate subthalamic nucleus: evidence for ordered but reversed body-map transformations from the primary motor cortex and the supplementary motor area. J Neurosci. 16(8):2671-83.
Newbold C, Richardson R, Huang CQ, Milojevic D, Cowan R, Shepherd R. An in vitro model for investigating impedance changes with cell growth and electrical stimulation: implications for cochlear implants. J Neural Eng. 2004 Dec;1(4):218-27.
Ni D, Shepherd RK, Seldon HL, Xu SA, Clark GM, Millard RE. Cochlear pathology following chronic electrical stimulation of the auditory nerve. I: Normal hearing kittens. Hear Res. 1992 Sep;62(1):63-81.
Nowak LG, Bullier J. Axons, but not cell bodies, are activated by electrical stimulation in cortical gray matter. I. Evidence from chronaxie measurements. Exp Brain Res 1998a;118:477–88.
Nowak LG, Bullier J. Axons, but not cell bodies, are activated by electrical stimulation in cortical gray matter. II. Evidence from selective inactivation of cell bodies and axon initial segments. Exp Brain Res 1998b;118:489–500.
Nowinski WL, Belov D, Pollak P, Benabid AL. Statistical analysis of 168 bilateral subthalamic nucleus implantations by means of the probabilistic functional atlas. Neurosurgery. 2005;57(4 Suppl):319-30.
Nowinski WL, Belov D. (2003) The Cerefy Neuroradiology Atlas: a Talairach-Tournoux atlas-based tool for analysis of neuroimages available over the internet. Neuroimage. 20(1):50-7.
Obeso JA, Olanow CW, Rodriguez-Oroz MC, Krack P, Kumar R, Lang AE. Deep-brain stimulation of the subthalamic nucleus or the pars interna of the globus pallidus in Parkinson's disease. N Engl J Med. 345:956-963, 2001.
Okun MS, Tagliati M, Pourfar M, Fernandez HH, Rodriguez RL, Alterman RL, Foote KD. Management of referred deep brain stimulation failures: a retrospective analysis from 2 movement disorders centers. Arch Neurol. 62(8):1250-5, 2005.
Oozeer M, Veraart C, Legat V, Delbeke J. Simulation of intra-orbital optic nerve electrical stimulation. Med Biol Eng Comput. 2005 Sep;43(5):608-17.
Otsuka T, Abe T, Tsukagawa T, Song WJ. Conductance-based model of the voltage-dependent generation of a plateau potential in subthalamic neurons. J Neurophysiol. 92(1):255-64, 2004.
Pagni CA, Altibrandi MG, Bentivoglio A, Caruso G, Cioni B, Fiorella C, Insola A, Lavano A, Maina R, Mazzone P, Signorelli CD, Sturiale C, Valzania F, Zeme S, Zenga F. Extradural motor cortex stimulation (EMCS) for Parkinson's disease. History and first results by the study group of the Italian neurosurgical society. Acta Neurochir Suppl. 2005; 93:113-9.
Parent A, Sato F, Wu Y, Gauthier J, Levesque M, Parent M. (2000) Organization of the basal ganglia: the importance of axonal collateralization. Trends Neurosci. 23(Supp 10):S20-S27.
Parent M, Lévesque M, Parent A. Two types of projection neurons in the internal pallidum of primates: single-axon tracing and three- dimensional reconstruction. J Comp Neurol. 439(2):162-75, 2001.
Parent M, Parent A. The pallidofugal motor fiber system in primates. Parkinsonism Relat Disord.10(4):203-11, 2004.
Patel NK, Heywood P, O'Sullivan K, McCarter R, Love S, Gill SS. Unilateral subthalamotomy in the treatment of Parkinson's disease. Brain. 2003; 126(Pt 5):1136-45.
Paxinos G, Huang XF, Toga AW. The Rhesus Monkey Brain in Stereotaxic Coordinates. Academic Press, 1999.
143 Percheron G, Lacourly N. [The lack of precision of thalamic stereotaxy based on Horsley and Clarke cranial coordinates in Macaca (author's transl)] Exp Brain Res., 1973; 18(4):355-73.
Perier C, Vila M, Feger J, Agid Y, Hirsch EC. Functional activity of zona incerta neurons is altered after nigrostriatal denervation in hemiparkinsonian rats. Exp Neurol. 2000;162(1):215-24.
Perlmutter JS, Mink JW, Bastian AJ, Zackowski K, Hershey T, Miyawaki E, Koller W, Videen TO. Blood flow responses to deep brain stimulation of thalamus. Neurology 2002;58:1388–94.
Plaha P, Ben-Shlomo Y, Patel NK, Gill SS. Stimulation of the caudal zona incerta is superior to stimulation of the subthalamic nucleus in improving contralateral parkinsonism. Brain. 2006;129(Pt 7):1732-47.
Plaha P, Gill SS. Bilateral deep brain stimulation of the pedunculopontine nucleus for Parkinson's disease. Neuroreport. 2005; 16(17):1883-7.
Plenz D, Kital ST. A basal ganglia pacemaker formed by the subthalamic nucleus and external globus pallidus. Nature. 1999;400(6745):677-82.
Poggi S, Pallotta S, Russo S, Gallina P, Torresin A, Bucciolini M. Neuronavigation accuracy dependence on CT and MR imaging parameters: a phantom-based study. Phys Med Biol., 2003; 48(14):2199-216.
Pralong E, Debatisse D, Maeder M, Vingerhoets F, Ghika J, Villemure JG. Effect of deep brain stimulation of GPI on neuronal activity of the thalamic nucleus ventralis oralis in a dystonic patient. Neurophysiol Clin. 2003;33(4):169-73.
Rall W, Burke RE, Holmes WR, Jack JJ, Redman SJ, Segev I. (1992) Matching dendritic neuron models to experimental data. Physiol Rev. 72(Supp 4):S159-S186.
Ranck JB Jr. Specific impedance of rabbit cerebral cortex. Exp Neurol. 7:144-52, 1963.
Ranck JB Jr. Which elements are excited in electrical stimulation of mammalian central nervous system: a review. Brain Res. 98(3):417-40, 1975.
Rattay F. Analysis of models for external stimulation of axons. IEEE Trans Biomed Eng 1986; 33:974-977.
Rattay F. The basic mechanism for the electrical stimulation of the nervous system. Neuroscience 1999; 89:335-346.
Rattay F, Minassian K, Dimitrijevic MR. Epidural electrical stimulation of posterior structures of the human lumbosacral cord: 2. quantitative analysis by computer modeling. Spinal Cord. 2000 Aug;38(8):473-89.
Rebert CS, Hurd RE, Matteucci MJ, De LaPaz R, Enzmann DR. A procedure for using proton magnetic resonance imaging to determine stereotaxic coordinates of the monkey's brain. J Neurosci Methods., 1991; 39(2):109-13.
Reichel M, Breyer T, Mayr W, Rattay F. Simulation of the three-dimensional electrical field in the course of functional electrical stimulation. Artif Organs. 2002 Mar;26(3):252-5.
Rizzone M, Lanotte M, Bergamasco B, Tavella A, Torre E, Faccani G, Melcarne A, Lopiano L. Deep brain stimulation of the subthalamic nucleus in Parkinson's disease: effects of variation in stimulation parameters. J Neurol Neurosurg Psychiatry. 71(2):215- 9, 2001.
Rodriguez-Oroz MC et al. Bilateral deep brain stimulation in Parkinson's disease: a multicentre study with 4 years follow-up. Brain. 128(Pt 10):2240-9, 2005.
Rohlfing T, Maurer CR Jr, Dean D, Maciunas RJ. Effect of changing patient position from supine to prone on the accuracy of a Brown-Roberts-Wells stereotactic head frame system. Neurosurgery., 2003; 52(3):610-8.
Rubin JE, Terman D. High frequency stimulation of the subthalamic nucleus eliminates pathological thalamic rhythmicity in a computational model. J Comput Neurosci. 2004;16(3):211-35.
Ruddy HA, Loeb GE. Influence of materials and geometry on fields produced by cochlear electrode arrays. Med Biol Eng Comput. 1995 Nov;33(6):793-801.
Saint-Cyr JA, Hoque T, Pereira LC, Dostrovsky JO, Hutchison WD, Mikulis DJ, Abosch A, Sime E, Lang AE, Lozano AM. Localization of clinically effective stimulating electrodes in the human subthalamic nucleus on magnetic resonance imaging. J Neurosurg. 2002;97(5):1152-66.
Salin P, Manrique C, Forni C, Kerkerian-Le Goff L. High-frequency stimulation of the subthalamic nucleus selectively reverses dopamine denervation-induced cellular defects in the output structures of the basal ganglia in the rat. J Neurosci 2002;15:5137–48.
Sato F, Parent M, Lévesque M, Parent A. Axonal branching pattern of neurons of the subthalamic nucleus in primates. J Comp Neurol. 424(1):142-52, 2000.
144
Saunders RC, Aigner TG, Frank JA. Magnetic resonance imaging of the rhesus monkey brain: use for stereotactic neurosurgery. Exp Brain Res., 1990; 81(2):443-6.
Schulte T, Brecht S, Herdegen T, Illert M, Mehdorn HM, Hamel W. Induction of immediate early gene expression by high-frequency stimulation of the subthalamic nucleus in rats. Neuroscience. 2006;138(4):1377-85.
Shen KZ, Zhu ZT, Munhall A, Johnson SW. (2003) Synaptic plasticity in rat subthalamic nucleus induced by high-frequency stimulation. Synapse. 50:314-319.
Shepherd RK, McCreery DB. Basis of electrical stimulation of the cochlea and the cochlear nucleus. Adv Otorhinolaryngol. 2006;64:186-205.
Shi LH, Luo F, Woodward DJ, Chang JY. Basal ganglia neural responses during behaviorally effective deep brain stimulation of the subthalamic nucleus in rats performing a treadmill locomotion test. Synapse. 2006;59(7):445-57.
Smith Y, Bolam JP, Von Krosigk M. Topographical and Synaptic Organization of the GABA-Containing Pallidosubthalamic Projection in the Rat. Eur J Neurosci. 2(6) :500-511, 1990.
Starr PA, Christine CW, Theodosopoulos PV, Lindsey N, Byrd D, Mosley A, Marks WJ Jr. Implantation of deep brain stimulators into the subthalamic nucleus: technical approach and magnetic resonance imaging-verified lead locations. J Neurosurg. 2002 Aug;97(2):370-87.
Stefani A, Fedele E, Galati S, Pepicelli O, Frasca S, Pierantozzi M, Peppe A, Brusa L, Orlacchio A, Hainsworth AH, Gattoni G, Stanzione P, Bernardi G, Raiteri M, Mazzone P. Subthalamic stimulation activates internal pallidus: evidence from cGMP microdialysis in PD patients. Ann Neurol. 2005;57(3):448-52.
Stefani A, Fedele E, Galati S, Raiteri M, Pepicelli O, Brusa L, Pierantozzi M, Peppe A, Pisani A, Gattoni G, Hainsworth AH, Bernardi G, Stanzione P, Mazzone P. Deep brain stimulation in Parkinson's disease patients: biochemical evidence. J Neural Transm Suppl. 2006;(70):401-8.
Strafella AP, Sadikot AF, Dagher A. Subthalamic deep brain stimulation does not induce striatal dopamine release in Parkinson's disease. Neuroreport. 2003;14(9):1287-9.
Struijk JJ, Holsheimer J, van Veen BK, Boom HB. Epidural spinal cord stimulation: calculation of field potentials with special reference to dorsal column nerve fibers. IEEE Trans Biomed Eng. 1991 Jan;38(1):104-10.
Sumanaweera TS, Adler JR Jr, Napel S, Glover GH. Characterization of spatial distortion in magnetic resonance imaging and its implications for stereotactic surgery. Neurosurgery., 1994; 35(4):696-703.
Swadlow HA. (1992) Monitoring the excitability of neocortical efferent neurons to direct activation by extracellular current pulses. J Neurophysiol. 68:605-19.
Sweeney JD, Ksienski DA, Mortimer JT. A nerve cuff technique for selective excitation of peripheral nerve trunk regions. IEEE Trans Biomed Eng. 1990 Jul;37(7):706-15.
Tagliati M, Shils J, Sun C, Alterman R. Deep brain stimulation for dystonia. Expert Rev Med Devices. 2004;1(1):33-41.
Tai CH, Boraud T, Bezard E, Bioulac B, Gross C, Benazzouz A. Electrophysiological and metabolic evidence that high-frequency stimulation of the subthalamic nucleus bridles neuronal activity in the subthalamic nucleus and the substantia nigra reticulata. FASEB J. 2003; 17(13):1820-30.
Tehovnik EJ. Electrical stimulation of neural tissue to evoke behavioral responses. J Neurosci Methods. 65(1):1-17, 1996.
Temperli P, Ghika J, Villemure JG, Burkhard PR, Bogousslavsky J, Vingerhoets FJ. How do parkinsonian signs return after discontinuation of subthalamic DBS? Neurology. 2003;60(1):78-81.
Terman D, Rubin JE, Yew AC, Wilson CJ. Activity patterns in a model for the subthalamopallidal network of the basal ganglia. J Neurosci. 22(7):2963-76, 2002.
Thobois S, Fraix V, Savasta M, Costes N, Pollak P, Mertens P, Koudsie A, Le Bars D, Benabid AL, Broussolle E. Chronic subthalamic nucleus stimulation and striatal D2 dopamine receptors in Parkinson's disease--A [(11)C]-raclopride PET study. J Neurol. 2003;250(10):1219-23.
Tuch DS, Wedeen VJ, Dale AM, George JS, Belliveau JW. (1999) Conductivity mapping of biological tissue using diffusion MRI. Ann N Y Acad Sci. 888:314-316.
Tuch DS, Wedeen VJ, Dale AM, George JS, Belliveau JW. (2001) Conductivity tensor mapping of the human brain using diffusion tensor MRI. Proc Natl Acad Sci. 98:11697-11701.
145 Turner RS, Anderson ME. Pallidal discharge related to the kinematics of reaching movements in two dimensions. J Neurophysiol., 1997; 77(3):1051-74.
Urbano FJ, Leznik E, Llinas RR. (2002) Cortical activation patterns evoked by afferent axons stimuli at different frequencies: an in vitro voltage-sensitive dye imaging study. Thalamus Rel Syst. 1:371-378.
Viola P, Wells WM. Alignment by maximization of mutual information. Int. J. Comput. Vision, 1997; 24(2):137-154.
Vitek JL, Giroux M. (2000) Physiology of hypokinetic and hyperkinetic movement disorders: model for dyskinesia. Ann Neurol 47(Suppl 1):S131-S140.
Vitek JL, Hashimoto T, Peoples J, DeLong MR, Bakay RA. Acute stimulation in the external segment of the globus pallidus improves parkinsonian motor signs. Mov Disord. 2004; 19(8):907-15.
Vitek JL. (2002a) Deep brain stimulation for Parkinson's disease. A critical re-evaluation of STN versus GPi DBS. Stereotact Funct Neurosurg. 2002;78(3-4):119-31.
Vitek JL. (2002b) Mechanisms of deep brain stimulation: excitation or inhibition. Mov Disord. 17(Supp 3):69-72.
Voges J, Volkmann J, Allert N, Lehrke R, Koulousakis A, Freund HJ, Sturm V. Bilateral high-frequency stimulation in the subthalamic nucleus for the treatment of Parkinson disease: correlation of therapeutic effect with anatomical electrode position. J Neurosurg. 2002;96(2):269-79.
Volkmann J, Herzog J, Kopper F, Deuschl G. Introduction to the programming of deep brain stimulators. Mov Disord. 17 Suppl 3:S181-7, 2002.
Walter BL, Vitek JL. Surgical treatment for Parkinson's disease. Lancet Neurol. 3(12):719-28, 2004.
Wang LY, Kaczmarek LK. (1998) High-frequency firing helps replenish the readily releasable pool of synaptic vesicles. Nature. 394:384-388.
Weiland JD, Anderson DJ. Chronic neural stimulation with thin-film, iridium oxide electrodes. IEEE Trans Biomed Eng. 2000 Jul;47(7):911-8.
Welter ML, Houeto JL, Bonnet AM, Bejjani PB, Mesnage V, Dormont D, Navarro S, Cornu P, Agid Y, Pidoux B. Effects of high- frequency stimulation on subthalamic neuronal activity in parkinsonian patients. Arch Neurol. 2004 Jan;61(1):89-96.
Wichmann T, Bergman H, DeLong MR. The primate subthalamic nucleus. I. Functional properties in intact animals. J Neurophysiol. 1994 Aug;72(2):494-506.
Wichmann T, Kliem MA, Soares J. Slow oscillatory discharge in the primate basal ganglia. J Neurophysiol. 87(2):1145-8, 2002.
Wilson CJ, Weyrick A, Terman D, Hallworth NE, Bevan MD. A model of reverse spike frequency adaptation and repetitive firing of subthalamic nucleus neurons. J Neurophysiol. 91(5):1963-80, 2004.
Windels F, Bruet N, Poupard A, Feuerstein C, Bertrand A, Savasta M. Influence of the frequency parameter on extracellular glutamate and gamma-aminobutyric acid in substantia nigra and globus pallidus during electrical stimulation of subthalamic nucleus in rats. J Neurosci Res 2003;72:259–67.
Windels F, Bruet N, Poupard A, Urbain N, Chouvet G, Feuerstein C, Savasta M. (2000) Effects of high frequency stimulation of subthalamic nucleus on extracellular glutamate and GABA in substantia nigra and globus pallidus in the normal rat. Eur J Neurosci. 12:4141-4146.
Windels F, Carcenac C, Poupard A, Savasta M. Pallidal origin of GABA release within the substantia nigra pars reticulata during high-frequency stimulation of the subthalamic nucleus. J Neurosci. 2005;25(20):5079-86. Wu YR, Levy R, Ashby P, Tasker RR, Dostrovsky JO. Does stimulation of the GPi control dyskinesia by activating inhibitory axons? Mov Disord 2001;16:208–16.
Yelnik J, Damier P, Demeret S, Gervais D, Bardinet E, Bejjani BP, Francois C, Houeto JL, Arnule I, Dormont D, Galanaud D, Pidoux B, Cornu P, Agid Y. Localization of stimulating electrodes in patients with Parkinson disease by using a three-dimensional atlas- magnetic resonance imaging coregistration method. J Neurosurg. 2003;99(1):89-99.
Yokoyama T, Ando N, Sugiyama K, Akamine S, Namba H. Relationship of stimulation site location within the subthalamic nucleus region to clinical effects on parkinsonian symptoms. Stereotact Funct Neurosurg. 2006;84(4):170-5.
Zonenshayn M, Sterio D, Kelly PJ, Rezai AR, Beric A. Location of the active contact within the subthalamic nucleus (STN) in the treatment of idiopathic Parkinson's disease. Surg Neurol. 2004;62(3):216-25.
Zucker RS, Regehr WG. (2002) Short-term synaptic plasticity. Annu Rev Physiol. 64:355-405.
146 APENDIX A: CICERONE USER MANUAL
147
3D visualization and database software for stereotaxic neurosurgical navigation in non-human primate research
Copyright © 2005
148 INSTALLATION
Cicerone was written using VTK (Visualization Toolkit; http://public.kitware.com/VTK/) and Tcl/Tk (Tool Command Language; http://tcl.sourceforge.net/). The current version runs on Windows. It should in theory run on any platform that has VTK with Tcl/Tk installed, but this has not yet been tested. To install Cicerone, run MonkeyCicerone_setup.exe. The setup wizard will create MonkeyCicerone folder on the C: drive (C:\MonkeyCicerone). If no setup executable was provided, copy MonkeyCicerone folder to C: drive. In this folder there will be MonkeyCicerone executable, Cicerone_DICOM_Sorter executable (see Part 3 of the manual), and five subfolders (\StandardAtlas, \chambers, \vta, \SampleData, \help). The StandardAtlas folder contains 3D brain nuclei. Folder ‘chambers’ contains geometries of head chambers and electrodes available for manipulation in Cicerone. Folder ‘vta’ contains pre-calculated estimates for volumes of tissue activated by a DBS electrode at various stimulation parameters. Folder SampleData contains sample MRI and CT images, three configuration files and recording data file. Folder ‘help’ contains help files that can be opened from within the program. The content of the help files is the same as this manual. Configuration file contains locations of imaging files and other setup variables, and it must be provided to Cicerone at the start of the program. A custom configuration file can be created by modifying the provided sample files (see Part 2). New configuration file can also be saved from within the program and it will reflect changes the user makes to the imaging volumes, 3D atlas, chambers and electrodes.
For questions, comments or bug reports please email Svjetlana Miocinovic ([email protected]) or Cameron McIntyre ([email protected]).
QUICK START
Using Sample Data
1. Start Cicerone, either by selecting it from the Start Menu or double-clicking on MonkeyCicerone.exe. At the window prompt that appears, click on Browse and chose a configuration file from SampleData folder. Two new windows will appear: a graphics window and a menu.
2. To plan chamber locations, first turn on the CT image by checking the CT box in the menu window. The image of a skull extracted from CT data will appear (this will take some time and will depend on the speed of the computer). To position chambers, click on ‘Chamber’ button on the right side of the menu. Up to three chambers can be placed, and they can be turned on and off by checking the box next to the appropriate ‘Chamber’ button. All distances are in millimeters and angles in degrees.
149 3. To mark electrode recording location, enter microdrive coordinates into the upper right section of the menu (this will position electrode within the chamber). Then click on button ‘Describe Site’ or press ‘s’. After pressing ‘Save’, a marker will appear at the tip of the electrode. Information about the recording site will be immediately saved to an output file Cicerone_Sites_date.txt, created in the output folder that was specified in the configuration file. To view atlas in 2D, click on ‘View by Track’ button and select a slice. You can also load in sample recording sites. To do this, click on ‘Read Data’ button and select Recording_sites.txt file from SampleData folder.
Using New Data
1. Prepare imaging files in DICOM or vtk format (see Part 3)
2. Write a new configuration file (see Part 2)
3. Start Cicerone and click on ‘Image Menu’. This opens a popup window with expanded functions for image manipulation.
4. Transform MRI so that its AC-PC line is horizontal and brain is symmetrical around the vertical axis. MRI’s AC should overlap with the large green sphere (Cicerone’s AC), and MRI’s PC should match up with the small green sphere (Cicerone’s PC). To do this, click on ‘Transform MRI’. Another window will appear with functions for MRI translation and rotation. Check ‘Grid’ box to view a grid of vertical and horizontal lines that can help align the image.
5. After MRI is properly placed, similar manipulations can be done for a CT image. Click on ‘Transform CT’ and rotate and translate CT so that it is coregistered with the MRI. To facilitate this process, check ‘Green-Red’ box; this will display MRI and CT in red and green colors. Then reduce CT transparency, turn on MRI slice and then a corresponding CT slice. If CT slice is turned on after MRI, it will appear transparent which will make lining up of the two slices easier.
6. Click on ‘Set Landmarks’ or ‘Transform CT’ (if using a CT), to set up ear bar and eye bars that define stereotactic coordinate space. Ear bar and eye bars should be positioned in the ear canals and on the inferior orbital ridges visible in the CT or MRI (if animal was scanned in a frame). During the chamber implantation surgery it is very important that the head be positioned in the same way in the actual stereotaxic frame as it is in Cicerone.
7. Scale the standard atlas to fit the animals MRI/CT (click on button ‘ALL’ to modify all nuclei at once, or click on individual nuclei buttons to modify each nucleus separately). The recommended procedure is to modify all the nuclei together, first scaling in the AP direction so that the atlas PC aligns with the small green sphere (MRI PC). Then scale the nuclei in ML and VD direction. Use ‘Clipping Planes’ function to cut 3D nuclei at the level of MRI slice in order to facilitate scaling process.
150 8. Change coordinate system to ‘Stereotaxic’ and position chambers.
Close Cicerone only by clicking the ‘Quit’ button (do not close the program by clicking on X in the window’s upper right corner because Cicerone may need to resave recording sites if any changes were made to them). Also, do not press ‘q’ or ‘e’ while in the rendering window because that will automatically exit the program. For more details keep reading…
USER MANUAL (Monkey Cicerone 1.0)
Monkey Cicerone is 3D visualization and database software for stereotaxic neurosurgical navigation in non-human primate research. The program displays co-registered MRI, CT and 3D brain nuclei volumes, along with head chambers and electrodes. Graphical user interface allows interactive visualization and manipulation of objects. Microelectrode recording site data can also be saved and displayed.
Cicerone can be used to: 1. Plan stereotaxic chamber placement surgery Up to 3 recording/stimulating chambers and electrodes (micro or DBS) can be positioned in the stereotaxic space. User can visualize electrode trajectory inside the brain and optimize chamber placement prior to surgery. Coordinates provided by Cicerone can be input directly into the stereotaxic frame (after the frame is zeroed). Chambers can be positioned in coronal, sagittal or any oblique orientation. 2. Visualize electrode trajectory during neurophysiologycal recording experiments Cicerone displays electrode position based on user input of microdrive coordinates. Markers are created for each recording site and user can view them interactively in 3D or 2D. User can also reposition 3D nuclei to better fit the recording data. 3. Visualize spread of deep brain stimulation (DBS) Theoretical prediction for volume of tissue activated (VTA) can be displayed for a given set of DBS parameters. The estimates were made for monkey DBS electrode (0.75 mm diameters, 0.5mm contacts with 0.5mm spacing). 4. Save recording sites from neurophysiologycal recording experiments Cicerone saves recording locations into a text file suitable for import into a database. Recording sites can also be uploaded from a text file into Cicerone.
The 3D brain atlas used in Cicerone was created from a brain atlas of longtailed macaque (M. fascicularis; Martin RF, Bowden DM. Primate Brain Maps: Structure of the Macaque Brain. Elsevier Science, 2000; http://braininfo.rprc.washington.edu). When appropriately scaled the atlas can be used in rhesus macaque (M. mulatta), Japanese macaque (M. fuscata) and baboon (P. papio, P. cynocephalus, P. anubis). In these species the scaled atlas accuracy has been evaluated only for the upper brain stem (diencephalon, midbrain and pons), so it may not be suitable for use in the cortex and lower brain stem.
151 For more information about Monkey Cicerone see: Miocinovic S, Zhang J, Xu W, Russo GS, Vitek JL, McIntyre CC. Stereotactic neurosurgical planning, recording, and visualization for deep brain stimulation in non-human primates. J Neurosci Methods. 2007 May 15;162(1-2):32-41.
This manual has five parts: 1. Part 1 - explains Cicerone's graphical user interface 2. Part 2 - explains how to create a configuration file 3. Part 3 - explains how to create MRI, CT and 3D nuclei files suitable for input into Cicerone. 4. Part 4 – contains practical tips 5. Appendix
PART 1: Graphical User Interface
When Cicerone is started, user will be prompted to enter configuration file name. Click on Browse to search for desired file. Configuration file can be just an empty text file. Next, two windows will open: one for the graphics display and one for the main menu.
MOUSE BUTTONS Left mouse button: Rotate the display Right mouse button: Zoom in/out Middle mouse button: Pan
Use Shift key if the mouse has only two buttons. Clicking on an object will display its name. Clicking on a recording site marker will open a popup window to revise the site information.
MAIN MENU
The main menu has several panels (Fig. A-1). Functions in each panel are explained below.
Panel A. MRI display Slide bars scroll through axial, sagittal and coronal MRI slices. Check boxes turn them on and off.
Panel B. Camera Position camera to view brain in the axial (transverse), sagittal, coronal or oblique orientation. The oblique view corresponds to the plane of the current chamber (see panel F). This function does not rotate imaging data; it only changes camera view.
Panel C. CT display Checkbox turns CT contour on and off (it will take several seconds to turn on). Contour value determines which contour from CT data will be displayed. Bone is usually best
152 seen around 500; soft tissue will appear at smaller values (~50-150), however this will depend on the range of the CT data. The slide bar changes CT transparency. Allow several seconds for display to update when changing CT properties. Click on ‘Image Menu’ button to view expanded imaging menu.
Panel D. Site Markers This panel allows user to turn recording site markers on or off. User can view them by Nucleus, Response, Track and Record. At each recording site, user can create a marker (a small sphere or another shape) and associate it with a specific nucleus, describe its response characteristics (motor and microstimulation) and indicate if neural firing was recorded at that site (note that Cicerone cannot record neural firing, but only its description). See also Panel E > ‘Describe Site’. Button ‘View by Track’ opens a pop-up window where user can: 1) Select which recording tracks are displayed (if there are more than 30 tracks, user can select which 30 are displayed) 2) Set depth calibration (see panel F) for each individual track. If global calibration is selected calibration defined in panel F will be used for all tracks. If variable calibration is selected, calibrations defined next to each track will be used. Click on ‘Update Site Marker Position’ to reposition site markers when changing calibration settings. 3) View atlas in 2D i.e. slice by slice (this function automatically adjusts clipping planes to display only a 0.25mm thick atlas slice; see panel k). Slice numbers refer to microdrive coordinates (e.g. slice 1 is where electrode is located when microdrive is set to 1mm medial or lateral (if chamber is sagittal or oblique) or 1mm anterior (if chamber is coronal or oblique)). If changing chamber position while this window is open, close and reopen the window to register the new chamber position.
Panel E. Data Storing
Describe Track A small window will appear where user can enter comments about the current recording track. Describe Site A popup window will appear where user can enter comments about the current recording site and assign it a location using a pull-down menu. User can describe neural response to passive/active movement and microstimulation if such exams were performed. Markers of different shapes will be created (sphere if no boxes are checked; cube if ‘arm’ is checked, cone if ‘leg’ is checked and cylinder if ‘other’ or more than one box is checked). When user clicks ‘Save’, data for the current recording site will be written to a text file (see section Output Files). User can leave site comment blank and location undetermined, but ‘Save’ must be clicked in order to save data to a file. Locations in the pull-down menu are set in the configuration file (site_markers) and can be modified by the user. See Table 4 for the list of available locations. Once a site is created, user can click on it and a window will open where site description can be modified or site deleted (if mouse click does not work, press ‘p’ while mouse is over a desired marker). When clicking ‘Save’ from the ‘Site Revision’ window, the changes will not be automatically written to the output text file. The user can either save
153
Figure 1. Cicerone main menu
the sites when exiting Cicerone (there will be an automatic prompt if any sites were modified) or save them by opening ‘View Stored Data’ and then clicking on ‘Save To File’ button.
Describe Recording A window will appear where user can enter comments about the current neural recording (record of neural activity saved using another program (e.g. Spike2); Cicerone 154 cannot save neural firing). Cicerone will generate a unique filename that can be used to name the record file. The filename is composed of the first two letters of the monkey’s name, current track, depth (decimal point is omitted) and a number in case one site has more than one recording associated with it (e.g. RH1_342_1). User must first save current site (‘Describe Site’) before a recording comment can be entered. When user clicks Save, recording comment will be written to a text file (see section Output Files).
View Stored Data A window will appear that shows a summary of stored data. Output data files contain more information than displayed in this window (see section Output Files). User can save the sites to a text file by clicking on ‘Save to File’. All sites that appear in this list will be saved.
Panel F. Electrode Positioning ‘Current Chamber’ radio buttons indicates which chamber’s electrode is currently being manipulated. ML, AP and Depth are coordinates that position electrode within the chamber; medial-lateral, anterior-posterior and depth, respectively. This is where the user enters microdrive coordinates to visualize electrode track during neurophysiological recording experiments. User should press Enter after a new value is entered. Positive AP coordinate moves the electrode anteriorly. Positive ML coordinate moves the electrode to the left; whether left corresponds to medial or lateral move, will depend on the chamber location. Depth should always be negative. Track indicates current recording track number. User should set the initial track number and increment it when starting a new track. Electrode indicates current electrode being used. User should increment this number when using a new recording electrode (to keep track of electrode changes). Calibration determines electrode tip location when microdrive is set to zero depth. User should measure the distance between the electrode tip and the upper chamber rim when the physical microdrive is set to zero depth, and enter this value as ‘Calibration’ (positive value if electrode tip is above the rim and negative if it is below). If Calibration is set to zero, tip of the electrode at zero depth is aligned with the top chamber rim. Every time the electrode is changed, it will not necessarily be zeroed to exactly the same position. To correct that, user can use variable calibration i.e. adjust each track’s depth separately (see Panel D; ‘View Sites by Track’).
Panel G. Brain Nuclei Modification Clicking on ‘Individual nuclei’ button will open a popup window with a list of 3D nuclei loaded in from the configuration file. Nuclei can be turned on/off by checking their corresponding box, and their transparency changed with the scale bar. Nuclei can be scaled and translated by clicking on their corresponding button. To transform all nuclei at once, click on ‘ALL’ button. This feature is used to customize the 3D atlas to animal’s MRI/CT, and to make adjustments during neurophysiological recordings (see Part 4). Nucleus can be translated or scaled in three directions: ML (medial-lateral), VD (ventral-dorsal) and AP (anterior-posterior). Translation and scaling values for ALL will be added to each individual nucleus. E.g. if STN nucleus is scaled by 1.2 in ML direction and ALL nuclei are scaled by -0.4 in ML direction, then the final STN scaling factor in the ML direction will be 0.8. ‘Return to Config’ button will return nucleus properties to those defined in the configuration file or to default (zero translation, scaling factor of 1, visibility
155 on, transparency 100) if nothing is specified in the configuration file. ‘Return To Last Saved’ will return nucleus to its position before user opened the pop-up window. ‘Cancel’ will return nucleus to its position before user opened the pop-up window and close the window. ‘Save’ will save current transformation and close the window. Visibility and transparency setting for ‘ALL’ takes precedence over individual nuclei settings, except for the initial display (when Cicerone starts up).
Panel H. Chamber Manipulation Up to 3 chambers and electrodes can be displayed in Cicerone. When planning chamber locations, coordinate system should be set to ‘Stereotaxic’ (see panel I). Checkboxes turn chambers and electrodes on and off. Clicking on a chamber button will open a pop-up window for chamber manipulation. Chamber angle determines chambers physical angle (shape). Chamber types available are: microrecording (round; 19mm inner diameter), DBS (round; 16mm inner diameter), cortical (square), Alpha-Omega (square; 27x27mm opening), and headpost (not technically a chamber, but it can be useful to plan its location as well). Two electrode types are: microrecording (0.1 mm diameter) and DBS (0.75mm diameter; contacts are 0.5 mm long and separated by 0.5mm). To set a chamber to sagittal orientation, Frame’s ML axis should be set to ‘perpendicular’, horizontal dial to ‘0’ and vertical dial to a desired angle (Fig. 2A). To set a chamber to coronal orientation, Frame’s ML axis should be set to ‘parallel’, horizontal dial to ‘-90’ and vertical dial to a desired angle (Fig. A-2B). Oblique orientation can be achieved by
Figure 2. Kopf Stereotaxic frame (model 1430). A) Frame’s ML axis is ‘perpendicular’ to the plane of rotation. B) Frame’s ML axis is ‘parallel’ to the plane of rotation. ‘Plane of rotation’ is the plane that the electrode holder defines when rotated back and forth on the vertical dial (rotation plane indicated by the red arrows).
156 moving horizontal dial between 0 and 90 (or 0 and -90). The actual stereotaxic frame should be set in the same way. Care should be taken so that oblique rotation and medial-lateral translation are properly applied to the physical frame. Positive and negative signs from Cicerone can be confusing because they depend on the chamber location (left or right side of the head), so the user should note if the actual translation/rotation is medial or lateral, and make sure that the same movement is applied to the actual frame. Cicerone chamber translation coordinates can be applied directly to the stereotaxic frame AFTER the physical frame is properly zeroed. This means that vertical and horizontal angles are set, and electrode holder stylet points to the center of the ear bars. Cicerone was designed for use with Kopf frame model 1430 and standard electrode holder model 1770 (Kopf Instruments, Tujunga, CA). Buttons on the bottom of the pop-up window reposition chamber to coordinates defined in the configuration file or to 0,0,0 if none were specified (‘Return to Config’); return the chamber to position before chamber pop-up window was opened (‘Return to Last Saved’); return the chamber to position before chamber pop-up window was opened and close the window (‘Cancel’); save current position and close the window (‘Save’). Recording sites are positioned with respect to the current chamber position. If the chamber is moved, sites can be repositioned by clicking on ‘Update Site Marker Position’. Chamber transformation coordinates are not modified when user changes coordinate system (see Panel I) i.e. they are defined with respect to the current coordinate system.
Panel I. Coordinate system User can choose between two coordinate systems: 1) AC-PC where anterior and posterior commissures are on a horizontal line and origin is set to the center of anterior commissure; 2) Stereotaxic (Hf0) where ear bar and eye bars are on a horizontal plane and origin is set to the intersection of the midline plane and the interaural line (i.e. middle of the earbar). The physical stereotaxic frame must be zeroed to the same origin (where the two ear bars meet i.e. the Frankfurt zero). AC-PC coordinate system is the default. MRI, CT and 3D nuclei are defined in AC-PC system, and this system must be selected when manipulating images. To position chambers, user should select Stereotaxic system. When system changes to Stereotaxic, Cicerone rotates the brain around the ML axis so that the earbar and eye bars are in-plane and shifts origin to the middle of the interaural line.
Panel J. Brain Atlas User can choose which side of the brain atlas to view (right, left, both, or none)
Panel K. Set Landmarks ‘Set Landmarks’ button opens up a pop-up window where user can position posterior commissure (PC), ear bar and eye bars. To use this feature the current coordinate system must be set to ‘AC-PC’. The same functions are provided in ‘Transform CT’ popup menu.
157 Clipping Planes Opens a pop-up window where user can set clipping planes so that only a section of the 3D atlas is visible. CT can only be clipped in the coronal anterior direction (this feature is useful to ‘see’ inside the skull).
Read Datafile Opens a pop-up window where user can specify name of a Cicerone data file (Cicerone_Sites.txt; see section Output Files for more details) to be opened. User can choose to append the new data to existing site markers or the replace the existing markers. New site markers will be positioned with respect to the current chamber, using ML, AP and Depth coordinates read in from the file.
View VTA Opens a pop-up window where user can turn on and visualize an estimate for the volume of tissue activated (VTA) by a DBS electrode for given stimulation parameters. These estimates were calculated using NEURON (http://www.neuron.yale.edu/neuron/) cable models of myelinated axons (2 um diameter) positioned perpendicular to the electrode shaft. The electrode was a scaled down version of a clinical DBS electrode (0.75 mm diameters, 0.5mm contacts with 0.5mm spacing). Homogeneous tissue conductance was assumed so VTAs will not change if electrode is moved to a different location.
Panel L. Save Configuration Saves a new configuration file that describes image transformations, landmark, chamber and 3D nuclei positions, and various other parameters (see Part 2). Clicking the button opens a pop-up window where user can choose a file name.
Save Picture Saves a picture of the display window in a tiff file. The image file is placed in C:\Cicerone directory (unless output_folder is defined in the configuration file).
Quit Exits the program. If any site descriptions have been altered, the user will be asked to save the data to a file.
IMAGE MENU
Image menu contains expanded set of functions for manipulating MRI and CT images. It can be accessed by clicking ‘Image Menu’ button in the main menu. The left side of the window is for MRI and the right side for CT manipulations (the second image does not have to be a CT; any volume data can be loaded). User can scroll through 3 orthogonal planes and turn them on/off. Image contrast can be set by adjusting Window and Level. ‘Landmarks’ checkbox turns on/off AC, PC, ear bar and eye bars. ‘Grid’
158 checkbox turns on/off a set of vertical and horizontal lines that can be used to help align the images. ‘Green-Red’ checkbox makes MRI red and CT green which is useful when aligning the two images (CT transparency should be set to ~50 and CT slice turned on after MRI to view the overlay). CT contour and transparency can be set just like in the main menu. Button ‘Transform MRI’ opens a window for MRI rotation and translation. Button ‘Transform CT’ opens a window for CT rotation and translation. User should first transform MRI and then CT. If CT image is not available, ‘Transform CT’ should still be opened to adjust location of ear and eye bars.
TRANSFORM MRI
It is necessary to align the MRI with Cicerone’s AC-PC coordinate system. The first step in this process is rotating the MRI so that coronal, sagittal and axial planes are correctly oriented. When user presses ‘Coronal View’ button, a coronal image should be displayed. On ‘Sagittal View’ a sagittal image should be displayed and anterior brain should be on the left side of the screen. Simple stated, in coronal view, animal’s head should be looking towards the screen. The landmarks (AC=big green sphere; PC=small green sphere; earbar = yellow cylinder; eye bars = yellow rectangles) can orient the user in the anterior- posterior direction. The 3D atlas (on the right by default) can help distinguish left and right. In case image is not oriented as described above, the user should press the three ‘Orient’ buttons until desired position is achieved. ‘Orient X’ rotates image 90 degrees around X axis (left-right axis); ‘Orient Y’ rotates image 90 degrees around Y axis (up-down axis); and ‘Orient Z’ rotates image 90 degrees around Z axis (front-back axis). On coronal view, if axial image is displayed, ‘Orient X’ should be pressed; if sagittal image is displayed, ‘Orient Y’ should be pressed. If on sagittal view, sagittal image is displayed but anterior brain is facing right, ‘Orient Y’ should be pressed twice. Next, user needs to determine if left and right side of the brain are displayed properly. It happens sometimes, especially with coronal DICOMs, that slices are put into a volume in the wrong order, so the left brain ends up on the right side and vice versa. If that is the case, user should flip the image by pressing ‘Reverse Slice Order’, and then redo step 1 (orienting the image). If there are no distinguishing characteristics in the image to distinguish left from right, the user should open the image in an image viewer (Cicerone_DICOM_Sorter can be used for DICOM files) and check what is left and right (viewers usually display right brain on the left side of the display window; same as Cicerone when ‘Coronal View’ is pressed). The third and final step in MRI transformation is aligning MRI’s AC and PC with Cicerone’s AC and PC landmarks (big and small green spheres). This is done by translating and rotating the MRI so that the brain is symmetrical around the vertical and horizontal axes (i.e. the brain is straight) and its AC-PC line is horizontal. Checking ‘Grid’ box will turn on a set of horizontal and vertical lines that can be used to help align the image. This process is done by sequentially moving through coronal, axial and sagittal view and rotating and translating image in the corresponding plane. Transformation plane is selected by choosing one of the radio buttons (axial, coronal and sagittal). Transformation distance (in
159 mm) or angle (in degrees) is selected with the next set of radio buttons (10, 5, 1, 0.5, 0.25). Position of the PC can be changed with ‘AC-PC distance’ scale bar. ‘Reset’ button will delete all transformations applied to the image – in that case user will need to start again from step 1. ‘Config File’ button puts image in a position defined in the configuration file (if no transformations were set in the configuration file, this is same as ‘Reset’).
TRANSFORM CT
Steps 1 and 2 for CT transformation are the same as for MRI. Step 3 is coregistering CT with the MRI. This is done be translating and rotating CT until it lines up with the MRI. To facilitate this process user can check ‘Green-Red’ checkbox which will turn MRI red and CT green so the two images can be more easily distinguished. Next, CT transparency should be set to ~50. If CT slice is turned on after a nearby MRI slice, the overlay of two images is visible. The final step is to position ear and eye bars which are the landmarks defining the stereotaxic space. This can be done either using a skull rendering extracted from the CT data or MRI if animal was scanned in a stereotactic frame. The view the skull, check CT contour box (it may take a few seconds to appear). Contour value can be adjusted until desired skull rendering is achieved. Ear bar should be placed to go through both ear canals (it should look symmetrical in axial view) and eye bars should rest on the bottom of the orbits. It may be necessary to further translate/rotate CT until landmarks fit properly (however, these changes should be small otherwise CT may no longer be coregistered with MRI).
OUTPUT FILES
Cicerone creates two output files: Cicerone_Sites_date.txt and Cicerone_Records_ date.txt, where ‘date’ is replaced by the actual date. If several files are saved on the same day, they are given number increments (e.g. Cicerone_Sites_date_2.txt). These files are created every time Cicerone is run in C:\Cicerone folder (unless output_folder is defined in the configuration file). Cicerone_Sites file contains information about each recording site that user saves (see main menu, Panel E, ‘Describe Site’). It is in a format suitable for import into Excel or Access (fields are separated by tabs). Fields included in the file for each saved recording site are listed in Table 1 (see Appendix). Cicerone_Records file contains descriptions of neural recordings. Cicerone does not record neural activity (this should be done using another program, e.g. Spike2), but it can save user comments about a recording at a specific site. It also creates a unique filename that can be used to save neural recording and associate it with Cicerone description. Example of such filename is RH5_256_1 where RH are the first two letters of a monkey’s name, 5 is the track number, 256 is the site depth (25.6mm); and 1 indicates this was the first recording for that site.
160 PART 2: Configuration file
Configuration file is a text file that contains all information necessary to load in animal’s MRI, CT and/or 3D brain nuclei image files, and position images, chambers and electrodes. A configuration file must be provided to Cicerone at the start of the program. To write configuration files user should use Notepad (not MSWord) so no unexpected formatting or extra characters are included in the file. The file should be saved as .txt file. There is a very specific syntax that the file requires, and user will not necessarily be warned if there are any errors. Forward slash (‘/’) must be used to separate directories, which is opposite of the Windows convention, and file names and directories should not contain any spaces. Start of a comment is indicated by ‘#’ character and everything after ‘#’ until the end of the line is ignored by Cicerone. A configuration file can be an empty text file; in that case only chambers and electrodes are displayed. See directory SampleData for examples of configuration files. When writing a configuration file, user would normally specify only: MonkeyName, MonkeyID, output_folder, mri_dicom_folder (or mri_filename), ct_dicom_folder (or ct_filename), site_markers and nuclei filenames. Everything else will be calculated and written out by Cicerone based on user input (see main menu, panel L, ‘Save Configuration’). Atlas nuclei that user wants to open should be specified as “abbreviation_filename file_pathname”, for example, Pu_filename C:/Cicerone/StandardAtlas/StandardAtlas_Pu.stl Table 3 lists nuclei that are available in Cicerone. New structures can be added to the atlas, however word 'ALL' is reserved and should not be used as a structure abbreviation. Also, abbreviation cannot start with a number. Every line of the configuration file should have 2 or 3 words (with a few exceptions). In case there are 2 words, the first word is Cicerone variable name and the second one is a value that the variable is assigned. If there are 3 words, the first one is array variable name, the second one is array index, and the third one is the variable value. Variable or array names cannot be changed by the user, only their values can be modified. However not all variables need to be included and they do not need to be defined in any specific order. See Table 2 for a list of configuration file variables. If a variable is not defined in a configuration file, it will be assigned a default value. The exceptions to 2/3 words per line rule are for variable site_markers and transformation matrices. Variables ending in mat_list describe transformation matrices and they are followed by 16 numbers. These numbers will generally not be input by user, but calculated by Cicerone. Variable site_markers can be followed by any number of locations from Table 4 (these are locations available in the pull-down menu in ‘Describe Site’ pop up window). User can also specify arbitrary location in site_markers list. The new location name must be a single word, and it will be assigned default color (beige). Location ‘other’ is included in site_markers by default.
161 PART 3: Cicerone Image Requirements
Cicerone can read in MRI and CT images in DICOM format and/or .vtk format (Visual ToolKit). Throughout the manual the first image volume was assumed to be an MRI and the second volume a CT, however any two image volumes in the appropriate format can be loaded. There are many different DICOM formats and Cicerone is not guaranteed to read all of them. DICOM files need to be raw (not encapsulated). Folder where DICOM files are located (mri_dicom_folder and ct_dicom_folder in configuration file) should only contain images from one brain volume. For example, if both T1 and T2 MRI scans were acquired, the scanner output software will likely place both volumes into the same folder. These two volumes must be separated before importing images into Cicerone. DICOM folder must not contain any localizer or similar extraneous images. If these requirements are not met, the program will likely crash or freeze. To help user with organizing DICOM files and determining if they can be read by Cicerone, a helper program called Cicerone_DICOM_Sorter is provided. In this program user can view DICOM images from a specified folder and select those images that form one brain volume. It is very important that localizer images, blank images or any other non-brain related images not be included. Images selected by the user can then be copied to C:/Cicerone/temp_dicom folder (user can move images to another folder before importing them into Cicerone). There are also many free DICOM image viewers and converters that can be used to prepare images for Cicerone (see http://www.idoimaging.com). Cicerone is not very reliable at making sure that the left and right side of the brain are displayed properly, so it is up to the user to determine if left and right need to be reversed (see ‘Reverse Slice Order’). This can be an issue when coronal DICOMs are imported; axial and sagittal images are usually read in correctly. Another, more comprehensive, imaging viewer could be used to determine what is left/right if this is not obvious from the image itself. The user can then apply a correction in Cicerone if necessary (see ‘Transform MRI’). Another option is to import images into Cicerone as .vtk volume files. Analyze (http://www.mayo.edu/bir/) or another imaging software can be used to load in images in a variety of formats. In Analyze, images can be saved as .hdr and .img files (Analyze format). Free program MicroView (http://microview.sourceforge.net/) converts Analyze .hdr and .img files into .vtk format, suitable for import into Cicerone. Cicerone cannot read DICOM files exported from Analyze. If two MRI volumes need to be fused into one (e.g. a coronal and a sagittal scan), this can be done in Analyze. In Analyze the images should be resampled to have the same voxel size in all directions and cut and/or padded to have the same number of voxels in each direction. They should then be coregistered (ITK>Registration). This will probably have to be done manually since automatic registration algorithm does not deal well with partial volumes. If images were acquired during the same scanning session, they will likely only need to be translated to coregister (unless the animal moved significantly in which case some rotation may need to be applied). The result should be saved as fused volume. The fused volume should then be saved in Analyze format (.hdr and .img) and converted to .vtk in Microview.
162 3D brain nuclei were created in graphics modeling program Rhinoceros (http://www.rhino3d.com) and saved in stl format (stereolithography; binary or ascii). Instead of using nuclei provided in the StandardAtlas, user can create custom nuclei. One option is to warp 2D atlas templates to MRI slices and create 3D volumes from warped cross-sections. Warping can be done in Edgewarp (http://vhp.med.umich.edu/edgewarpss.html) or another program. In Rhino, nuclei should be saved in coronal orientation facing forward. All stl objects are automatically rotated by 90 degrees around the x axis because Cicerone and Rhino use different coordinate systems. Chambers and electrodes were also created in Rhino, in stl format. Their filenames are hardcoded into Cicerone (see Chambers/ folder) and cannot be changed by the user. If requested, different chamber and electrode shapes can be provided. User could also put a different stl object under one of the existing filenames. For example, in Chambers/ folder there are two folders called withcap/ and withoutcap/ containing chambers with and without caps. To display chambers with caps, copy files from withcap/ to Chambers/ (replace existing files). To go back to displaying chambers without caps, copy files from withoutcap/ to Chambers/ (replace existing files).
PART 4: Practical Tips
RUNNING CICERONE First try running Cicerone using provided sample configuration files. If that works, you can move on to making your own configuration file (see Part 2). If Cicerone does not run with sample data, there could be a few potential reasons: 1. Open configuration file and make sure that all the files really do exist in their specified locations. If they are not where they should be, put them there. 2. If Cicerone seems to open, but then crashes or freezes, it could be that the graphics card cannot support it. Make sure that you have the latest card driver. 3. If nothing works, try installing Cicerone on a different (newer) computer.
IMAGING DATA Cicerone can work with only a limited set of imaging data and provides limited tools for image manipulation. It will take some patience and practice to get everything set up correctly. Please read carefully part 3, and part 1 sections: ‘Image Menu’, ‘Transform MRI’, and ‘Transform CT’.
CHAMBER PLANNING If during surgery the chambers do not fit like they did in Cicerone, it is most likely that monkey’s head is not positioned like it was in Cicerone (it is very easy to misplace the earbars). Alternatively, there could be some error in applying Cicerone coordinates to the frame, but this should have been checked for before the surgery. Make sure that monkey’s head is placed firmly within the frame and that both ear and eye bars are in their proper
163 location. Palate bar should be positioned correctly so that the head does not rotate around the ear bars. Also when planning chamber locations, don’t put them too close (within a couple of millimeters) because it is likely that there will be some discrepancies between the virtual head and real head position, so the chambers may not fit exactly as planned. It is also advisable to get a CT scan after chamber implantation to verify their position. This CT can be imported into Cicerone and virtual chamber locations adjusted accordingly.
MICROELECTRODE RECORDINGS Cicerone provides a 3D atlas that can be used as guidance during microelectrode recordings. It is likely that the recordings do not fit the 3D atlas exactly. There could be a few reasons: 1. The scaled atlas is not quite right. Make adjustments so that the atlas fits animal’s MRI as well as possible. 2. There was a brain shift after the craniotomy or monkey’s head is in a different position. Both of these reasons will cause the brain to be in a different position with respect to the skull than it was during imaging. For example, if MRI is acquired with monkey in a supine position and microrecordings are done in a sitting position, it may be necessary to move the 3D atlas anteriorly since the brain likely shifted anteriorly within the skull (1.5 mm shift was necessary for the three of our monkeys) 3. The recording chamber is not where you think it is (this can be solved by doing a CT after chamber implantation) 4. The microdrive is misaligned so electrode is not in the expected location, or the electrode is deflected, so it is not following the perfectly straight trajectory that Cicerone assumes.
Since there will always be some discrepancies, you can move 3D atlas to better match your microrecordings data (click on ‘ALL’ button in Modify Nuclei section of the main menu). The movement should be on the order of couple of millimeters – anything more suggests that problems 1-4 may be occurring. Moving the atlas after the first few tracks, usually makes the subsequent tracks fit better. Also note that tracks may be slightly offset from each other in depth if electrodes were not zeroed properly. To fix that, adjust ‘variable depth calibration’ (see ‘View Sites by Track’).
164 APPENDIX (Tables 1-4)
Table 1. Fields included in Cicerone_Sites output file Field Description Data Type SiteNumber Site number (starts with 1) integer TrackNumber Track number (set by user) integer AP AP electrode (microdrive) coordinate number ML ML electrode (microdrive) coordinate number Chamber Recording chamber (1,2 or 3) 1, 2 or 3 MonkeyID Monkey ID text MonkeyName Monkey name text TrackComment Track comment text Date Date text Depth Depth electrode (microdrive) negative number coordinate Location Site location (nucleus) from Table 4 or user defined SiteComment Site comment text MotorResponse Presence of neural response to any combination of passive or active movement of arm, words: arm leg other leg or other MicrostimResponse Presence of response to any combination of microstimulation words: arm leg other RecordFile Recording Filename one or more words ElectrodeCalibr Electrode calibration number (usually negative) ElectrodeNumber Electrode used (set by user) integer
165
Table 2. Cicerone configuration file variables Variable Name Value Default Description Comment type value MonkeyName string “” monkey name MonkeyID string “” monkey ID output_folder string C:/Monk folder where output Use full pathname eyCicero files will be written (no spaces) ne mri_dicom_folder string folder containing MRI Use full pathname DICOM files (no spaces) ct_dicom_folder string folder containing CT Use full pathname DICOM files (no spaces) mri_filename string MRI vtk file Use full pathname (no spaces) ct_filename string CT vtk file Use full pathname (no spaces) nucleus*_filename string 3D nucleus stl file Use full pathname (no spaces) site_markers locations other locations that recording can be followed by from Table sites can be assigned to any number of 4 (in ‘Describe Site’) locations from Table 4 orient_mat_list 16 1 0 0 0 0 MRI orientation matrix numbers 1 0 0 0 0 1 0 0 0 0 1 mri_mat_list 16 1 0 0 0 0 MRI translation matrix numbers 1 0 0 0 0 1 0 0 0 0 1 reslice_mat_list 16 1 0 0 0 0 MRI rotation numbers 1 0 0 0 0 matrix 1 0 0 0 0 1 ct_orient_mat_lis 16 1 0 0 0 0 CT orientation matrix t numbers 1 0 0 0 0 1 0 0 0 0 1 ct_mat_list 16 1 0 0 0 0 CT translation matrix numbers 1 0 0 0 0 1 0 0 0 0 1 ct_reslice_mat_lis 16 1 0 0 0 0 CT rotation
166 t numbers 1 0 0 0 0 matrix 1 0 0 0 0 1 mri_flip_flag 1 or 0 0 1 if MRI slices need to be reordered; 0 otherwise ct_flip_flag 1 or 0 0 1 if CT slices need to be reordered; 0 otherwise ct_contour number (CT data value of the contour Bone is usually within CT range) / extracted from CT data best seen at 500 data range 7 to visualize the skull if data ranges from -1000to3000 landmarks number -11 AC-PC distance Can be set in pc_z_trans from -30 (negative number) ‘Transform MRI’ to 0 in 0.1 menu increments landmarks number -14 Earbar VD distance Can be set in earbar_y_trans from -50 from AC-PC plane ‘Transform CT’ to 100 in (negative number) menu 0.1 increments landmarks number -20 Earbar AP distance Can be set in earbar_z_trans from -50 from AC ‘Transform CT’ to 100 in (negative number) menu 0.1 increments landmarks number 70 Earbar length Can be set in earbar_length from 30 ‘Transform CT’ to 100 in 1 menu increments landmarks number 17 Left orbitbar ML Can be set in orbitbar_x_trans from -50 distance from AC ‘Transform CT’ to 100 in menu 0.1 increments landmarks number -15 Left orbitbar VD Can be set in orbitbar_y_trans from -50 distance from AC-PC ‘Transform CT’ to 100 in plane (negative number) menu 0.1 increments landmarks number 31 Earbar AP distance Can be set in orbitbar_z_trans from -50 from AC ‘Transform CT’ to 100 in (positive number) menu 0.1 increments landmarks number 35 Distance between left Can be set in 167 orbitbar_distance from 0 to and right orbit bars ‘Transform CT’ 50 in 0.1 menu increments current_system acpc acpc AC-PC or or Stereotaxic Frankfurt stereoHf0 Zero chview1 ** 1 or 0 1 for Chamber1 visibility (1 is chamber on, 0 is off) 1, 0 for chamber s 2 and 3 changle1 0 – 60 in 5 0 Chamber1 physical angle degree (determines its shape) increments chtype1 micro, dbs, micro Chamber1 cortex, chamber type aomega, or headpost chelec1 Micro or micro Chamber1 dbs electrode type chtrans x1 number 0 Chamber1 Medial- from - Lateral (ML) translation 100 to 100 in mm in 0.1 increments chtrans y1 number 0 Chamber1 Ventral- from - Dorsal (VD) translation 100 to 100 in mm in 0.1 increments chtrans z1 number 0 Chamber1 Anterior- from - Posterior (AP) 100 to 100 translation in mm in 0.1 increments chtrans rx1 number 0 Chamber1 vertical dial from - 180 to 180 in 0.5 increments chtrans ry1 number 0 Chamber1 horizontal dial from - 180 to 180 in 0.5 increments chtrans rz1 number 0 not used 168 from - 180 to 180 in 0.5 increments chtrans rcy1 number 0 Chamber1 rotation to fit from - the skull 180 to 180 in 0.5 increments chtrans plane1 0 or 1 0 Frame’s ML axis is parallel (1) or perpendicular (0) to plane of rotation eltrans x1 number 0 Chamber1 electrode ML This is the ML from -10 translation within the microdrive to 10 in 0.1 chamber (microdrive coordinate that increments coordinates) in mm user would input eltrans y1 number 0 Chamber1 electrode VD This is the depth from - (depth) translation microdrive 100 to 100 within the chamber coordinate that in 0.1 (microdrive coordinates) user would input increments in mm eltrans z1 number 0 Chamber1 electrode AP This is the AP from -10 translation within the microdrive to 10 in 0.1 chamber (microdrive coordinate that increments coordinates) in mm user would input calibration elec1 number 0 Chamber1 electrode Distance of calibration electrode tip to top chamber rim at zero depth (positive if electrode tip is above chamber, negative if below) nuctrans number 0 nucleus Medial-Lateral x_all is added to x_nucleus *** from -5 (ML) translation in mm x_ for individual to 5 in nuclei 0.05 increments nuctrans y_ number 0 nucleus Ventral-Dorsal y_all is added to nucleus from -5 (VD) translation in mm y_ for individual to 5 in nuclei 0.05 increments nuctrans z_ number 0 nucleus Anterior- z_all is added to nucleus from -5 Posterior (AP) z_ for individual
169 to 5 in translation in mm nuclei 0.05 increments nuctrans sx_ number 1 (0 for nucleus Medial-Lateral sx_all is added to nucleus from 0.1 sx_all) (ML) scaling sx_ for individual (-1 for nuclei sx_all) to 2 in 0.02 increments nuctrans sy_ number 1 (0 for nucleus Ventral-Dorsal sy_all is added to nucleus from 0.1 sy_all) (VD) scaling sy_ for individual (-1 for nuclei sy_all) to 2 in 0.02 increments nuctrans sz_ number 1 (0 for nucleus Anterior- sz_all is added to nucleus from 0.1 sz_all) Posterior (AP) scaling sz_ for individual (-1 for nuclei sz_all) to 2 in 0.02 increments nuctrans view_ 1 or 0 1 nucleus visibility (1 is view_nucleus nucleus on, 0 is off) takes precedence over view_all during the initial display. Then view_all gets priority. nuctrans opacity_ number 100 nucleus transparency opacity_nucleus nucleus from 0 to (100 means completely takes precedence 100 in 1 non transparent) over opacity_all increments during the initial display. Then opacity_all gets priority. * nucleus can be any nucleus from Table 3. ** There are 3 electrodes and 3 chambers so all ‘chtrans’, ‘eltrans’, ‘chview’, ‘chtype’, ‘changle’ and ‘calibration elec’ variables are repeated for chamber2 and chamber3. *** nucleus can be any nucleus from Table 3 or ‘all’ is which case transformation coordinates are applied to all nuclei.
170 Table 3. Cicerone 3D atlas nuclei. Abbreviations are from Martin and Bowden atlas. ATLAS ABBR DESCRIPTION COMMENT approximate structure - sulci and Cx cerebral cortex gyri not accurately represented from 8ac to -12ac; approximate structure - sulci and gyri not mCx middle portion of cerebral cortex accurately represented last few posterior slices not cw cerebral white matter included includes hippocampus (Hi), dentate gyrus (DG), alveus, (alv), fimbria of HiF HIPPOCAMPAL FORMATION hippocampus (fi), and subiculum (S) includes body, genu and splenium of cc corpus callosum corpus callosum approximate structure - some curves connected to create icap internal capsule continuous structure ac anterior commissure includes part of OLV (occipital LV lateral ventricle horn of lateral ventricle) includes part of OLV (occipital TLV temporal horn of lateral ventricle horn of lateral ventricle) Cd caudate nucleus includes head, body and tail Pu putamen LGP lateral globus pallidus MGP medial globus pallidus lml lateral medullary lamina includes corticomedial and basolateral nuclear group (CoA, MeA, LOT, CeA, BL, LA, BA, ABA, Amg amygdala and PA (from -3ac to 2ac)) includes dorsal, lateral, medial and triangular septal nuclei (DS, LS, S SEPTAL NUCLEI MS, TS) includes anterior column, body and fx fornix posterior column of fornix Cl claustrum SI substantia inominata Acb nucleus accumbens olf olfactory tract st stria terminalis includes lateral and medial Hb HABENULA habenular nuclei (LHb, MHb) Pi pineal body
171 sm stria medullaris of thalamus also included as part of thalamus includes thalamic nuclei, thalamic fiber tracts and stria medullaris of Th THALAMUS thalamus Th_AD anterodorsal nucleus Th_AM anteromedial nucleus Th_AV anteroventral nucleus Th_PT paratenial nucleus paraventricular nucleus of Th_PV thalamus Th_Re reuniens nucleus Th_SF subfascicular nucleus includes paralaminar, margocellular and parvicellular parts of MD Th_MD MEDIAL DORSAL NUCLEUS (MDPL, MDM, MDPC) Th_MDPL paralaminar part of MD Th_MDM magnocellular part of MD Th_MDPC parvicellular part of MD Th_CD central dorsal nucleus Th_CL central lateral nucleus Th_CM central medial nucleus Th_PC paracentral nucleus Th_CMn centromedian nucleus Th_PF parafascicular nucleus Th_LD lateral dorsal nucleus Th_LP lateral posterior nucleus includes oral, lateral, medial and inferior pulvinar nuclei (OPul, LPul, Th_Pul PULVINAR MPul, IPul) Th_OPul oral pulvinar nucleus Th_LPul lateral pulvinar nucleus Th_MPul medial pulvinar nucleus Th_IPul inferior pulvinar nucleus Th_VAPC parvicellular part of VA Th_VAMC magnocellular part of VA Th_VLO oral part of VL Th_VLC caudal part of VL Th_VLM medial part of VL Th_VLP pars posterna of VL Th_VPLO oral part of VPL Th_VPLC caudal part of VPL VENTRAL POSTEROMEDIAL includes principal and parvicellular Th_VPM NUCLEUS parts of VPM (VPMPr and VPMPC) Th_VPMPr principal part of VPM
172 Th_VPMPC parvicellular part of VPM dorsal nucleus of lateral Th_DLG geniculate body ventral nucleus of lateral Th_VLG geniculate body includes ventral and magnocellular nuclei and capsule of MGB (VMG, Th_MGB MEDIAL GENICULATE BODY MMG, cmg) Th_Lim limitans nucleus Th_SG suprageniculate nucleus Th_SM submedial nucleus Th_Rt thalamic reticular nucleus ANTERIOR HYPOTHALAMIC AHR REGION INTERMEDIATE IHR HYPOTHALAMIC REGION POSTERIOR HYPOTHALAMIC PHR REGION LATERAL HYPOTHALAMIC LH AREA TM tuberomammillary nucleus DH dorsal hypothalamic area STh subthalamic nucleus ZI zona incerta H field H H1 field H1 H2 field H2 al ansa lenticularis thirdV third ventricle tdf telo-diencephalic fissure includes optic tract (opt) and optic opt optic tract chiasm (ox) Ptec pretectal region SC SUPERIOR COLLICULUS IC INFERIOR COLLICULUS pc posterior commissure MIDBRAIN RETICULAR MBRF FORMATION CnF cuneiform nucleus pedunculopontine tegmental PPTg nucleus includes parvocellular and magnocellular part, and capsule of R RED NUCLEUS red nucleus (RPC, RMC, cr)
173 Aq cerebral aqueduct central gray substance of CGMB midbrain DR dorsal raphe nucleus VTA ventral tegmental area SNC substantia nigra pars compacta SNR substantia nigra pars reticulata ccr cerebral crus includes superior central nucleus (SuC), oral and pontine reticular PONTINE RETICULAR nuclei (PnO, PnC), and PRF FORMATION reticulotegmental nucleus (RtTg) scp superior cerebellar peduncle mcp middle cerebellar peduncle Cb cerebellum approximate structure Bs lower brainstem approximate structure
174 Table 4. Site marker locations that have assigned display color. Other location names can be specified by user Site Description Display Color striatum striatum tan putamen putamen red caudate caudate banana stn subthalamic nucleus raspberry zi zona incerta ultramarine gpi globus pallidus pars interna (MGP) blue gpe globus pallidus pars externa (LGP) green sn substantia nigra, unspecified plum snr substantia nigra pars reticulata lavender snc substantia nigra pars compacta grey thalamus thalamus, unspecified aquamarine thal_VLO thalamic VLO yellow_ochre thal_VPLO thalamic VPLO cobalt_violet_deep
thal_VLC thalamic VLC maroon
thal_VPLC thalamic VPLC tomato thal_Retic thalamic reticular nucleus royal_blue optictract optic tract purple cortex cortex turquoise background location where background signal goes yellow up or down quiet location without neural activity black
border border cell chocolate fiber location with possible fiber activity lemon_chiffon
other unspecified seashell
175