Journal of Neuroscience Methods 149 (2005) 154–163

Method of construction of a MRI-based tabular databaseof 3D stereotaxic co-ordinates for individual structures

in the basal ganglia ofMacaca mulatta

Milind Deogaonkara,∗, Marcel Heersa, Supriya Mahajana,Marijn Brummerb, Thyagarajan Subramaniana

a Department of Neuroscience, Mailcode NB20, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USAb Department of Radiology, Emory University School of Medicine, Atlanta, GA, USA

Received 25 September 2004; received in revised form 10 May 2005; accepted 16 May 2005

Abstract

Primate models are commonly used in Parkinson’s disease research to study stereotaxic strategies that demand accurate localization of thes es of variousb ach animal,t in the axialp outlineso t reliablet present ther f this studyi tially add toa aging havem tion basedh duce the costsc©

K

1

rmeBKtn

rgeturo-dis-

rimatei-ls,do-

et

al.,

0d

tructures in basal ganglia. We demonstrate a method to construct an extensive tabular database of 3D stereotaxic co-ordinatasal ganglia structures from high-quality magnetic resonance (MR) images of 47 adult female 3–5 kg rhesus monkeys. For e

he structures in the basal ganglia were traced as they appeared on the axial MR images. Their maximal outlines were projectedlane to create a stack of images andX, Y, Z co-ordinates were calculated for margins of each structure. These co-ordinates and thef the individual nuclei help delineate a “common area,” which was further narrowed down to a point that represents the ‘mos

arget point’ (MRTP) in subthalamic nucleus, globus pallidum, caudate and putamen on both sides. Common area and MRTP reegion that can most definitely be associated with a structure and hence the most definite target for a given structure. The goal os to demonstrate the method of construction, discuss the feasibility and usefulness of such a tabular database that could potenccuracy of localization while using atlas-based stereotaxy. Though use of MRI remains a standard practice and advances in imade targeting for functional surgery more accurate, in developing countries that implies prohibitive costs per procedure. Populauman databases similar to the monkey database described here, when used along with less expensive imaging modalities can reonsiderably as well as add to the accuracy of targeting.2005 Elsevier B.V. All rights reserved.

eywords: MRI; Macaca mulatta; Stereotaxy; Basal ganglia; Parkinson’s disease

. Introduction

To understand the pathophysiology of parkinsonism andelated disorders and develop therapeutic approaches, pri-ate models are commonly used. These functional mod-ls closely resemble human disorders (Bodis-Wollner, 1990;oyce et al., 1990; Close et al., 1990; Guttman et al., 1990;aakkola and Teravainen, 1990). Macaca mulatta in par-

icular has been extensively used as a functional model ofeurodegenerative disorders.

∗ Corresponding author. Tel.: +1 216 4452358; fax: +1 216 445 1466.E-mail address: deoganm@ccf.org (M. Deogaonkar).

Stereotaxy is used in these animals primarily to tavarious nuclei for neuroanatomical, neurochemical, nephysiological and neurosurgical studies. In Parkinson’sease, for example, stereotaxic approaches are used in pmodels to study the pathophysiology (Allen et al., 1986; DMonte et al., 2000; Langston et al., 2000), the effect of antiparkinsonian drugs (Akai et al., 1993; Andringa and Coo2000; Andringa et al., 2003; Savola et al., 2003), understanthe neuroanatomical correlates (Bergman et al., 1994; Thedosopoulos et al., 2003), the effects of lesioning (Krack etal., 2002) and stimulation (Hashimoto et al., 2003; Zabekal., 2003) and to develop novel transplantation (Bakay andHerring, 1989; Lindvall and Bjorklund, 1989; Bakay et

165-0270/$ – see front matter © 2005 Elsevier B.V. All rights reserved.oi:10.1016/j.jneumeth.2005.05.016

M. Deogaonkar et al. / Journal of Neuroscience Methods 149 (2005) 154–163 155

1998) and gene therapy strategies (Bjorklund et al., 2000;Kordower et al., 2000). Stereotaxic strategies are also usedin electrophysiological studies (Raz et al., 2000).

Stereotaxic procedures in monkeys are often done usingatlas-based coordinates. However, a previous study by ourgroup reported that this approach can lead to inaccura-cies in targeting due to insufficient stereotaxic precision(Subramanian et al., 2001; Subramanian et al., 2005). Ourfindings were similar to that of a study by Percheron(Percheron, 1975). MRI guidance improves the accuracyof targeting various structures in basal ganglia in primates(Subramanian et al., 2005). Electrophysiology guidance alsoimproves the targeting of deeper and smaller structures likesubstantia nigra (Starr et al., 2000) and subthalamic nucleus(Cuny et al., 2002) in humans and non-human primates.Moreover, the current atlas ofM. mulatta is based on a singleanimal study (Paxinos, 1999) and does not take into accountindividual variations within a species. This can lead to errorsin targeting due to individual variations. To overcome thisproblem efforts have been made in the past by constructinga digital template atlas that provides a template to comparebrain structures in individuals for internal variability in mon-keys (Martin and Bowden, 1996; Black et al., 2001; Greer etal., 2002) as well as in humans (J Mazziota et al., 1995; Togaand Thompson, 2001).

The main goal of the present study was to create an exten-s cturei fromh iplea ures.I sim-p reep e theC ll thea hisd es int , andn thet pri-m od toi achi thist log-i fora erapyl

2

2

w andw malsw are

and Use of Laboratory Animals’, in stainless steel cages inair-conditioned and temperature and light-cycle-controlledrooms. All procedures were carried out in strict compliancewith the “Principles of laboratory animal care” (NationalInstitute of Health Publication No. 86-23, revised 1985) andwere approved by the institute’s animal care and use com-mittee.

2.2. MRI and data acquisition

Prior to the MRI scans, animals were anaesthetized usingintravenous Propofol (2,6-diisopropylphenol) infusion, theanimal were then placed in a standard large animal Kopfstereotaxy frame made with Plexiglas and non-ferromagneticmetal parts. The plastic ear bars of the head frame were filledwith MR contrast agent so they are visible on the MR images.Two additional MR contrast filled posts were placed anteriorat the level of eye bars (eye-posts) so that a plane drawnbetween the anterior posts and ear bars will be parallel to thesidebars of stereotaxic frame. MRI images were then obtainedin a 1.5 Tesla Philips Gyroscan NT scanner (Philips MedicalSystems, Cleveland, OH). Several series of axial, coronal,and sagittal scout images were acquired in order to determinethe plane intersecting the four gadolinium-enhanced fiducialmarkers. This plane scouting procedure implicitly providesstereotaxic registration at scan time. T1-weighted gradiente ereo ng onalt xcita-t eT wo wasa SunU ew,C agea tors.I erser 3Dv sianc rmatp lineso hei ion.T erveda alidityo withM tingt Them rawnt turesw iplea MRs idualn aten

ive database of 3D stereotaxic co-ordinates for each strun the basal ganglia using external reference pointsigh-quality magnetic resonance (MR) images from multnimals and to precisely chart the location of these struct

n this study we do not morph the images and use ale technique to visualize each individual structure in therpendicular planes, outline the boundaries, determinartesian co-ordinates, and overlap the images from animals to derive a mean for that individual structure. Tatabase can be used to identify various brain structur

he basal ganglia, in neuroanatomical, neurochemicaleurophysiological studies and for guiding probes into

arget nuclei during stereotaxic surgery in nonhumanates. Using this technique we describe a novel meth

dentify a common geometric location in the middle of endividual basal ganglia structure. Initial targeting usingechnique can be further refined with additional physiocal targeting techniques like microelectrode recordingsccurate placement of probes, cell transplants or gene th

oci.

. Material and methods

.1. Animals

Forty-seven adult female Rhesus macaques (M. mulatta)ere used for this study. Animals were of 3 to 5 kg weightere obtained from IACUC approved sources. The aniere housed as recommended in NIH ‘Guide for the C

cho 3D images with isotropic 0.7 mm resolution wbtained on a 256× 256× 100 acquisition/reconstructiorid, reconstructed in coronal plane orientation orthog

o this marker plane. Scan parameters included an eion angleα = 30◦, repetition time TR = 28 ms, echo timE = 4 mswith NEX = 2 on an in-plane square field of vief size 160 mm. The total imaging time for this scanpproximately 30 min. The images were transferred to altra10 workstation (Sun Microsystems, Mountain ViA) for pre-operative analysis using the BrainWork imnalysis software, developed at Emory by the investiga

nteractively, orthogonal coronal, sagittal, and transveformatted image planes were generated from theoxel volume. The BrainWork software provides Carteo-ordinates at the intersection point of the three refolanes which are displayed as cross-hair referenceverlaying the images (Fig. 1). Stereotaxic coordinates of t

ndividual brain structures were determined in this fashhe plane going through the eye-posts and ear-bars ss a standardized zero reference plane to ensure the vf the coordinates between individuals. Ear-bars filledR contrast were used as zero point while calcula

he dorso-ventral and antero-posterior co-ordinates.edio-lateral zero was at the point where the plane d

hrough the ear-bars meets the midline. These strucere then overlapped with the images from all multnimals to determine the mean values. Consecutiveections were stacked together to reconstruct the indivuclei of the basal ganglia like globus pallidum, cauducleus, putamen and subthalamic nucleus (Figs. 2 and 3).

156 M. Deogaonkar et al. / Journal of Neuroscience Methods 149 (2005) 154–163

Fig. 1. T1-weighted MR image showing the reformatting of brain images used to determine theX, Y, Z co-ordinates. TheX, Y, Z co-ordinates were determinedby using intrinsic three-axes spatial co-ordinates (red, blue and green lines) and an external reference plane (ear-bar).

A 3D plot of the outlines of these major nuclei wasplotted.

2.3. Construction of tabular database and calculation of‘common area’ and ‘most reliable target points’ for eachstructure

Mean co-ordinates for the anterior, posterior, superior,and inferior margins of each structure were derived forthe individual structures. This was done by calculatingthe intrinsicX, Y, Z co-ordinates for various points on theanterior, posterior, medial and lateral margins of individualstructures and then correcting it for the ear-bar co-ordinateswhich was used as an external reference point (zero-point forcalculating the dorso-ventral, antero-posterior co-ordinateswith medio-lateral zero being at the point where plane drawnfrom ear-bars meet the midline). An example for the caudatenucleus of one of the monkeys is illustrated inTable 1.

Around 500 such points were charted on anterior, posterior,medial and lateral boundaries of each individual structureusing images from different contrast settings (by changingthe TR and TE on the images using BrainWork imageanalysis software) so as to reveal exact boundaries of eachstructure. Data tables were constructed and mean area andstandard deviation (S.D.) for each structure was calculated(Table 2). Once the mean co-ordinates for the margins wereavailable, we derived ‘common area’ by calculating theglobal mean for all the boundaries of a particular structure(Table 2). We also derived co-ordinates for multiple pointsinside the individual nuclei. The global mean of stereotaxicco-ordinates of multiple points from multiple animals ofan individual nucleus in the basal ganglia gave us the‘most reliable target point’ (MRTP) for that particularstructure (Table 2). Sigmaplot was then used to reconstruct3D images from the mean co-ordinates obtained from thedatabase.

M.D

eogaonkaretal./JournalofN

euroscienceM

ethods149

(2005)154–163

157

Table 1A tabular database of the intrinsicX, Y, Z co-ordinates for various points on the anterior, posterior, medial and lateral margins of individual structures corrected using the ear-bar co-ordinates which has beenused as an external reference point

Monkey: 0049 tpw scan 50

Caudate nucleus (left) Earbars:−x, −7.66,−8.59 Caudate nucleus (right) AdjustedX AdjustedY AdjustedZ

X Y Z AdjustedX AdjustedY AdjustedZ X Y Z

Superior 6.41 8.91 5.78 −6.41 16.57 14.37 −6.09 6.09 5 78 6.09 13.75 14.37Anterior 6.72 14.22 5.47 −6.72 21.88 14.06 −7.03 15.47 5.47 7.03 23.13 14.06Medial 4.22 11.72 5.47 −4.22 19.38 14.06 −3.28 10.47 5.47 3.23 18.13 14.06Posterior 6.72 1.41 5.47 −6.72 9.07 14.06 −6.41 2.34 5.47 6.41 10 14.06Lateral 8.59 10.16 5.47 −8.59 17.82 14.06 −7.97 5.16 5.47 7.97 12.82 14.06Anterior 6.72 14.22 5.16 −6.72 21.88 13.75 −7.03 15.47 5.16 7.03 23.13 13.75Medial 4.22 11.72 5.16 −4.22 19.38 13.75 −3.28 10.47 5.16 3.23 18.13 13.75Posterior 6.72 1.41 5.16 −6.72 9.07 13.75 −6.41 2.34 5.16 6.41 10 13.75Lateral 8.59 10.16 5.16 −8.59 17.82 13.75 −7.97 5.16 5.16 7.97 12.82 13.75Anterior 7.03 15.16 4.84 −7.03 22.82 13 43 −7.34 15.47 4.84 7.34 23.13 13.43Medial 2.03 8.91 4.84 −2.03 16.57 13.43 −2.97 4.84 4.84 2.97 12.5 13.43Posterior 6.72 −0.47 4.84 −6.72 7.19 13.43 −6.72 −0.47 4.84 6.72 7.19 13.43Lateral 8.28 12.34 4.84 −8.28 20 13.43 −8.28 11.09 4.84 8.23 18.75 13.43

The data here shows raw and corrected (using ear-bar point as zero point) co-ordinates for representative points collected for left and right caudatenucleus in one monkey.

158M

.Deogaonkar

etal./JournalofNeuroscience

Methods

149(2005)

154–163

Table 2An example of methodology used in calculating the ‘common area’ and ‘MRTP’ using the tabular database

Monkey: 00 Audate nucleus (right) Caudate nucleus (right)

AdjustedX AdjustedY

AdjustedZ

Z level fromear-bars

AdjustedX

AdjustedY

AdjustedZ

Z level fromear-bars

AdjustedX

AdjustedY

AdjustedZ

Z level fromear-bars

AdjustedX

AdjustedY

AdjustedZ

Z level fromear-bars

Anterior 7.34 23.43 18.44 22 7.03 23.13 14.06 22 7.03 20.31 16.25 19 6.38 28.75 16.25 10Medial 5.47 16.25 18.44 22 3.28 13.13 14.06 22 3.59 16.25 18.25 19 3.75 25.62 18.25 10Posterior 5.47 16.25 18.44 22 6.41 10 14.06 22 6.72 8.43 16.25 19 6.25 15.62 16.25 10Lateral 922 19.37 18.44 22 7.97 12.82 14.06 22 8.59 19.06 16.25 19 8.75 26.25 16.25 10Anterior 7.34 23.43 18.13 21 7.03 23.13 13.75 21 7.03 20.31 15.93 18 6.88 29.37 15.62 9Medial 547 16.25 18.13 21 3.28 18.13 13.75 21 3.59 16.25 15.93 18 3.13 24.37 15.62 9Posterior 5.47 16.25 18.13 21 6.41 10 13.75 21 6.72 8.43 15.93 18 6.25 12.5 15.62 9Lateral 9.22 19.37 18 13 21 7.97 12.82 13.75 21 8.59 19.06 15.93 18 8.75 24.37 15.62 9Anterior 6.72 25.31 17.81 20 7.34 23.13 13.43 20 6.09 20.93 15.62 17 6.25 30 15 8Medial 2.97 21.56 17.81 20 2.97 12.5 13.43 20 2.66 15 15.62 17 3.13 26.37 15 8Posterior 6.72 13.43 17.81 20 6.72 7.19 13.43 20 7.66 6.87 15.62 17 3.13 8.75 15 8Lateral 9.84 19.37 17.81 20 8.28 18.75 13.43 20 8.91 19.06 15.62 17 9.38 25.65 15 8Anterior 6.72 25.31 17.5 19 7.34 23.13 13.12 19 6.09 20.93 15.31 16 6.38 31.25 14.37 7Medial 2.97 21.56 17.5 19 2.97 12.5 13.12 19 2.66 15 15.31 16 2,5 23.75 14.37 7Posterior 6.72 13.43 17.5 19 6.72 7.19 13.12 19 7.66 6.87 15.31 16 6.25 12.5 14.37 7Lateral 9.84 19.37 17.5 19 8.28 18.75 13.12 19 8.91 19.06 15.31 16 9.38 26.87 14.37 7

Means Standard deviation Mean of means

X Y Z S.D.X S.D.Y S.D.Z X Y Z

13.646667 10.416667 5.16 7.434624 13.44127 9.5508237 12.329323 12.041979 5 Most reliable target point12.97 12.9225 5.16 6.7657323 7.0221667 9.550323710.3125 12.2975 5.16 8.056531 7.3786285 9.550823710.4725 11.595 5.16 6.506117 7.996518 9.5508237 Standard deviation of the means

10.4725 11.205 5.16 5.9718639 8.4215616 9.3750698 X Y Z

11.41 11.0325 5.16 4.8742658 7.9084823 9.3750698 1.4366462 1.0764827 0.165247312.5025 11.8925 5.16 5.0152526 7.0625178 9.375069813.1275 12.2825 5.16 5.1201537 6.7386492 9.3750698 Points at plus and minus each S.D.15.315 13.6875 4.84 6.7584934 6.5428199 9.1937769 13.765969 13.118462 5.165247312.2675 11.875 4.84 7.1691765 6.3852982 9.1937769 13.765969 10.965497 5.165247310.86 9.53 4.84 5.1526886 9.0171651 9.1937769 13.765969 13.118462 4.834752714.14 12.7325 4.84 5.6455587 6.0772767 9.1937769 13.765969 10.965497 4.834752713.2825 12.89 4.84 6.9707263 6.0718751 8.0259511 10.892677 13.118462 5.165247312.8925 12.875 4.84 7.9320001 5.7275969 8.0259511 10.892677 10.965497 5.165247312.11 12.875 4.84 8.8081969 5.7275969 8.0259511 10.892677 13.118462 4.834752711.4875 12.5625 4 84 8.6938345 5.6862546 8.0259511 10.892677 10.965497 4.8347527

Minimum of 500 such points charted on each monkey leading to calculating the mean, standard deviation and then mean of means. The global mean of means was used as the MRTP. The data shown inTable 1is further expanded and analyzed adding more points from various areas of a particular structure (caudate nucleus in this particular table). Up to 500points are collected for each structure in each monkey. Amean of all these points is derived. A global mean from multiple animals then gives the MRTP along with the standard deviation.

M. Deogaonkar et al. / Journal of Neuroscience Methods 149 (2005) 154–163 159

Fig. 2. Reformatted magnetic resonance image with stacking of outlinedportion of putamen (P) and caudate head (C).

Fig. 3. A 3D reconstruction of the globus pallidus using stacking of sequen-tial magnetic resonance images from an axial T1-weighted MRI scan.

3. Results

3.1. Sagittal and axial projections of Striatum, Pallidum,thalamus and subthalamic nucleus in Macaca mulatta

Considerable inter-animal anatomic differences wereobserved in the placement, size, and outline of the struc-

Fig. 4. XY scatter diagram outlining the GPe (left hemisphere) boundariesfrom magnetic resonance images.

tures in the basal ganglia. For each animal, the structures inthe basal ganglia were traced as they appeared on the axialimages. Their maximal outlines were projected in the axialplane to create a stack of images that showed the 3D outline ofthe structures (Figs. 2 and 3). The outlines constructed usingstereotaxy co-ordinates showed the characteristic shape ofthe nuclei as shown inFigs. 4–6. The globus pallidus exter-nus (GPe) inFig. 4 was lenticular in shape and oriented inan antero-medial to postero-lateral direction. The thalamusand subthalamic nucleus (STN) inFig. 5 also conformedto the shape, dimensions and orientation as seen in the MRimages. When superimposed, the same structure from differ-ent animals showed a varying degree of overlapping (Fig. 6).This suggests that the degree of variability changes depend-ing on the structure and margin (anterior, posterior, medialor lateral) and so does the error region. As an example, theerror region in left putamen is the lateral border of the puta-men as it showed the largest variations among the animals(Fig. 6).

3.2. Determination and validation of ‘common area’and ‘most reliable target point’ (MRTP)

Projecting the outlines of the individual nuclei fromall the animals allowed us to delineate a “common area”(Figs. 6 and 7). Fig. 7 shows the reconstruction of the GPeu -p turef o allt riveds als rea.”T pointtT s of− [(11

sing the MR derived co-ordinates, andFig. 6shows overlaping of putamen from multiple animals. When same struc

rom multiple animals overlap, the area that is common the animals can be derived as “common area”. When detatistically the meanX, Y, Z co-ordinates for that individutructure also confirm to the presence of this “common ahe common area was further narrowed down to a target

hat represented the “MRTP” in that structure (Figs. 6 and 7).he MRTP for the left GPe, for example, had co-ordinate8.7,−2.49 and 11.77. The ‘most reliable target point’X

S.D.),Y (S.D.),Z (S.D.)] for left caudate was−3.49 (±2.11),2.93 (±5.26), 9.03 (±3.90); right caudate 7.82 (±3.94),2.59 (±3.65), 3.66 (±5.84); left putamen−9.57 (±4.20),

160 M. Deogaonkar et al. / Journal of Neuroscience Methods 149 (2005) 154–163

Fig. 5. 3D plot of the thalamus and subthalamic nucleus (STN) in a single animal. It shows the anatomical accuracy of the co-ordinates in delineating theboundaries and regions of various nuclei.

10.78 (±5.94), 8.34 (±3.09); right putamen 10.43 (±3.09),10.96 (±3.66), 10.22 (±3.46). The MRTP values of variousstructures in basal ganglia calculated using the database aregiven inTable 3. The large standard deviation suggests thatthere is considerable inter-animal difference. The higher theS.D., more are the chances of targeting error.Fig. 6thus doc-uments the error region based on high S.D. on the marginsof the putamen. TheX, Y, Z co-ordinates for MRTP shown inTable 3are the intrinsic co-ordinates for that point corrected

using the ear-bar co-ordinates as the zero-point which hasbeen used as an external reference point.

Once the MRTP was determined for a particular structure itwas validated by randomly applying it to various MRI scans.For example, the ‘MRTP’ for left caudate was−3.49 (±2.11),12.93 (±5.26), 9.03 (±3.90). When this point was chartedafter ear-bar correction in MRI scans of 10 different animalsnot used in calculating this tabular database, the point alwayswas located in the left caudate.

Table 3A tabular database for MRTP for various structures in basal ganglia and the standard deviation

Structure MRTP co-ordinates (CorrectedX, Y, Z co-ordinatesusing ear-bars as zero reference point)

Standard deviation forX, Y, Zcoordinates of MRTP (±)

Caudate nucleus left −3.49, 12.93, 9.03 2.11, 5.26, 3.90Caudate nucleus right 7.82, 12.59, 3.66 3.94, 3.65, 5.84Globus pallidus extermus, left −8.7,−2.5, 11.77 2.85, 2.7, 2Globus pallidus externus, right 8.02,−2.5, 11.77 3.0, 2.6, 2.1Globus pallidus internus, left −8.4,−6.1, 12.1 1.8, 1.8, 3.3Globus pallidus internus, right 7.43,−6.17, 12 l.9, 1.7, 3.35Putamen, left −9.57, 10.78, 8.34 4.20, 5.94, 3.09Putamkn, right 10.43, 10.96, 10.22 3.09, 3.66, 3.46Thalamus, left −6.05,−9.30, 14.87 2.73, 3.51, 2.15Thalamus, right 3.49,−8.87, 15.13 2.48, 3.47, 2.06Subthalamic nucleus, left −3.47,−6.76, 9.21 1.64, 1.57, l.12Subthalamic nucleus, left 3.4l,−6.98, 9.39 1.5, 1.59, 1.23

The co-ordinates given for MRTP are corrected co-ordinates according to the ear-bar zero as the external reference point.

M. Deogaonkar et al. / Journal of Neuroscience Methods 149 (2005) 154–163 161

Fig. 6. 3D plot of boundaries of the putamen usingX, Y, Z co-ordinates from multiple animals showing the overlapped area and MRTP. The central point isMRTP while the 4 peripheral points are in the region of high S.D. and thus represent the error region.

Fig. 7. 3D plot of the globus pallidus externus (GPe) boundaries usingX,Y, Z co-ordinates from consecutive MRI slices outlining the ‘common area’for the GPe.

The practical implication of this common or overlappingarea is that it represents the region that can most definitelybe associated with that particular structure during stereotaxy.Identifying the “MRTP” further improves the accuracy oftargeting. Accuracy of stereotaxy depends on correcting themean dimensions of a target according to the actual dimen-sion of the individual. The overlapping common area andthe mean co-ordinates for each individual structure helpedus determine the most dependable way to reach any giventarget. We averaged these “MRTP” and “common areas” tocalculate the most definite target for a given structure, whicheliminates the inter-subject variability.

4. Discussion

Stereotaxy is used as a tool in diverse areas of neuroscienceresearch. The accuracy of stereotaxy depends on how wellthe user can determine the actual position of target in a givenbrain. MR guided stereotaxy is the most accurate method oftargeting (Subramanian et al., 2001, 2005). Errors in target-ing when using an atlas are generally a result of morphologicdifferences between individual macaque brains (Percheron,

162 M. Deogaonkar et al. / Journal of Neuroscience Methods 149 (2005) 154–163

1975; Subramanian et al., 2001) or human error in identify-ing the brain landmarks. Current atlases are either entirelybased on a single animal or do not have adequate 3D data(Olzewski, 1952; Paxinos, 1999). This shortcoming in theexisting atlases was rightly pointed out by various previousstudies (J Mazziota et al., 1995; Martin and Bowden, 1996;Toga and Thompson, 2001).

In this study, we focus on the methodology to create anextensive MR based multi-subject 3D database of variousstructures in basal ganglia ofM. mulatta that will compli-ment accuracy while using atlas based targeting. We havecalculated these co-ordinates using an external reference sys-tem. The zero reference plane used is the plane going throughthe eye-posts and ear-bars and it provides a rigid referenceplane. As same reference planes are used in primate stereo-taxy the ‘common area’ and the ‘MRTP’ that we have definedfor each structure should be applicable to any given animal.We have derived these points as a mean of 3D stereotaxicco-ordinates from 47 animals, which will account for theinter-animal variability. There are previous studies that haveused a similar rationale, but our database differs from themin a variety of aspects. The probabilistic atlas of human brainuses segmentation of the MRI images and uses computa-tional methods to deform and warp the data in a complexnon-linear deformation (J Mazziota et al., 1995). We havenot warped the images but used them to derive Cartesian co-o doesn stillg sidea pinga a thatr truc-t raina orka ag-i likebi rgeryt ntalp henu rrec-t gliam m-m rivedf thec issueo andfi taxicc n thei tionc tlasea rnaln s orvw ref-e 8

using advanced computational protocols and linear transfor-mation, individual structures can get blurred due to spatialvariability (Toga and Thompson, 2001). Since our databaseaccounts for this spatial variability by statistically derivinga “common area” and ‘MRTP’ (Figs. 6 and 7) the need forvisual identification of a particular structure is eliminated.The foremost advantage of this reconstituted database is thatit is an MRI-based 3D representation of various structuresin basal ganglia ofM. mulatta brain using multiple animalimages represented in a numerical form without using com-plicated techniques such as deforming. This method can beapplied to various other intra-cerebral structures apart fromthe ones that we illustrate in this study. This method can alsobe applied to other animal brains. It can, in future also serveas a basis for incorporating similar data gathered from otherimaging or electrophysiological modalities. One of the limi-tations of our study is use of 3–5 kg female animals. Validityof any population-based database is limited to the popula-tion similar to that used for data acquisition. We do not knowthe validity of this technique for the animals that are outsidethe age and sex we used. Additional studies in animals ofdifferent age, sex and weight are needed to overcome thislimitation.

In summary, we demonstrate a method of constructing MRimages based 3D stereotaxic co-ordinates database for var-ious structures in basal ganglia ofM. mulatta. This methodd rm-i areac agingt ouldp tereo-t ncesi orea itivec es liket dali-t to thea

A

n bef .2

R

A rtialkeys.

A P, ettmentani-7.

rdinate using external reference points. This methodot involve any complex computational modeling butives co-ordinates for the outlines and multiple points inparticular structure in the basal ganglia. This overlap

nd averaging of these points provides a common areepresents the most reliable location of that particular sure in any given animal. The template atlas of Macaque blso uses a similar rationale but is different from this ws it uses only two animals, ventriculography as the im

ng technique and essentially internal reference pointsicommisural plane (Martin and Bowden, 1996). Though

nternal reference points are commonly used in neurosuoday it obligates imaging. MR imaging in the experimerimates is at times logistically difficult. Our database wsed along with pre-existing atlases with necessary co

ion in the target point in a particular structure in basal ganay alleviate the need of imaging. The ‘MRTP’ and “coon areas” for each structure in our database are de

rom MR images. This inherently gives it an edge overo-ordinates derived from histological atlases due to thef distortion and shrinkage of tissue during perfusionxation. This method of creating a database of stereoo-ordinates for each structure taking into considerationter-animal variability recognized by the standard deviaan circumvent these in-accuracies. MR images based at times have low resolution and may not show the inteuclei of basal ganglia unless volume rendering algorithmoxel-based co-registration is used (Black et al., 2001). Evenhen multiple MRI scans are mapped on a pre-existingrence system like Talairach (Talairach and Tournoux, 198)

s

oes not involve complex computational methods of defong or warping the images. The MRTP and commono-ordinates in this database are arrived at after averhe MR images from multiple animals. This database crove to be a useful adjunct when using atlas based s

axy to enhance the accuracy and reliability. Though advan imaging have made targeting for functional surgery mccurate, in developing countries that implies prohibosts per procedure. Population based human databashis when used along with less expensive imaging moies can reduce the costs considerably as well as addccuracy of targeting.

ppendix A. Supplementary data

Supplementary data associated with this article caound, in the online version, atdoi:10.1016/j.jneumeth005.05.016.

eferences

kai T, Yamaguchi M, Mizuta E, Kuno S. Effects of terguride, a paD2 agonist, on MPTP-lesioned parkinsonian cynomolgus monAnn Neurol 1993;33:507–11.

llen JM, Cross AJ, Yeats JC, Ghatei MA, McGregor GP, Close Sal. Neuropeptides and dopamine in the marmoset. Effect of treawith 1-methyl-4-phenyl-1, 2, 3, 6 tetrahydropyridine (MPTP): anmal model for Parkinson’s disease. Brain 1986;109(Pt 1):143–5

M. Deogaonkar et al. / Journal of Neuroscience Methods 149 (2005) 154–163 163

Andringa G, Cools AR. The neuroprotective effects of CGP 3466B inthe best in vivo model of Parkinson’s disease, the bilaterally MPTP-treated rhesus monkey. J Neural Transm Suppl 2000:215–25.

Andringa G, Eshuis S, Perentes E, Maguire RP, Roth D, Ibrahim M, et al.TCH346 prevents motor symptoms and loss of striatal FDOPA uptakein bilaterally MPTP-treated primates. Neurobiol Dis 2003;14:205–17.

Bakay RA, Herring CJ. Central nervous system grafting in the treatmentof parkinsonism. Stereotact Funct Neurosurg 1989;53:1–20.

Bakay RA, Boyer KL, Freed CR, Ansari AA. Immunological responsesto injury and grafting in the central nervous system of nonhumanprimates. Cell Transplant 1998;7:109–20.

Bergman H, Wichmann T, Karmon B, DeLong MR. The primate subthala-mic nucleus II. Neuronal activity in the MPTP model of parkinsonism.J Neurophysiol 1994;72:507–20.

Bjorklund A, Kirik D, Rosenblad C, Georgievska B, Lundberg C, MandelRJ. Towards a neuroprotective gene therapy for Parkinson’s disease:use of adenovirus, AAV and lentivirus vectors for gene transfer ofGDNF to the nigrostriatal system in the rat Parkinson model. BrainRes 2000;886:82–98.

Black KJ, Koller JM, Snyder AZ, Perlmutter JS. Template imagesfor nonhuman primate neuroimaging: 2 Macaque. Neuroimage2001;14:744–8.

Bodis-Wollner I. Visual deficits related to dopamine deficiency in exper-imental animals and Parkinson’s disease patients. Trends Neurosci1990;13:296–302.

Boyce S, Rupniak NM, Steventon MJ, Iversen SD. Characterisation ofdyskinesias induced by L-dopa in MPTP-treated squirrel monkeys.Psychopharmacology (Berl) 1990;102:21–7.

Close SP, Elliott PJ, Hayes AG, Marriott AS. Effects of classical and novelagents in a MPTP-induced reversible model of Parkinson’s disease.Psychopharmacology (Berl) 1990;102:295–300.

C k oftisticalgical

D M,rkin-sord

G thising.

G hyl-andm. J

H ofrons.

J f theage

Kaakkola S, Teravainen H. Animal models of parkinsonism. PharmacolToxicol 1990;67:95–100.

Kordower JH, Emborg ME, Bloch J, Ma SY, Chu Y, Leventhal L,et al. Neurodegeneration prevented by lentiviral vector delivery ofGDNF in primate models of Parkinson’s disease. Science 2000;290:767–73.

Krack P, Dostrovsky J, Ilinsky I, Kultas-Ilinsky K, Lenz F, Lozano A, etal. Surgery of the motor thalamus: problems with the present nomen-clatures. Mov Disord 2002;17(Suppl 3):S2–8.

Langston JW, Quik M, Petzinger G, Jakowec M, Di Monte DA. Investi-gating levodopa-induced dyskinesias in the parkinsonian primate. AnnNeurol 2000;47:S79–89.

Lindvall O, Bjorklund A. Transplantation strategies in the treatment ofParkinson’s disease: experimental basis and clinical trials. Acta NeurolScand Suppl 1989;126:197–210.

Martin RF, Bowden DM. A stereotaxic template atlas of the macaquebrain for digital imaging and quantitative neuroanatomy. Neuroimage1996;4:119–50.

Olzewski J. The thalamus ofMacaca mulatta An atlas for use with thestereotaxic instruments. Basel, New York: Karger; 1952.

Paxinos G. In: Rhesus monkey brain in steretaxic co-ordinates. First Edi-tion Academic Press; 1999.

Percheron G. Ventricular landmarks for thalamic stereotaxy in Macaca. JMed Primatol 1975;4:217–44.

Raz A, Vaadia E, Bergman H. Firing patterns and correlations of sponta-neous discharge of pallidal neurons in the normal and the tremulous1-methyl-4-phenyl-1,2,3, 6-tetrahydropyridine vervet model of parkin-sonism. J Neurosci 2000;20:8559–71.

Savola JM, Hill M, Engstrom M, Merivuori H, Wurster S, McGuire SG, etal. Fipamezole (JP-1730) is a potent alpha2 adrenergic receptor antag-onist that reduces levodopa-induced dyskinesia in the MPTP-lesioned

872–

S gicalman

S RIplan-tracts

S nceand

2–80.T brain.

T s ofarkin-

T –53.Z of

son’s–447.

uny E, Guehl D, Burbaud P, Gross C, Dousset V, Rougier A. Lacagreement between direct magnetic resonance imaging and stadetermination of a subthalamic target: the role of electrophysiologuidance. J Neurosurg 2002;97:591–7.

i Monte DA, McCormack A, Petzinger G, Janson AM, QuikLangston WJ. Relationship among nigrostriatal denervation, pasonism, and dyskinesias in the MPTP primate model. Mov Di2000;15:459–66.

reer PJ, Villemagne VL, Ruszkiewicz J, Graves AK, Meltzer CC, MaCA, et al. MR atlas of the baboon brain for functional neuroimagBrain Res Bull 2002;58:429–38.

uttman M, Fibiger HC, Jakubovic A, Calne DB. Intracarotid 1-met4-phenyl-1,2,3,6-tetrahydropyridine administration: biochemicalbehavioral observations in a primate model of hemiparkinsonisNeurochem 1990;54:1329–34.

ashimoto T, Elder CM, Okun MS, Patrick SK, Vitek JL. Stimulationthe subthalamic nucleus changes the firing pattern of pallidal neuJ Neurosci 2003;23:1916–23.

Mazziota AT, Evans A, Fox P, Lancaster J. A probabalistic atlas ohuman brain: Theory and rationale for its development. Neuroim1995;2:89–101.

primate model of Parkinson’s disease. Mov Disord 2003;18:83.

tarr PA, Subramanian T, Bakay RA, Wichmann T. Electrophysiololocalization of the substantia nigra in the parkinsonian nonhuprimate. J Neurosurg 2000;93:704–10.

ubramanian T, Venkiteswaran K, Marchionini D, Brummer B. Mguidance improves accuracy of stereotactic targeting of cell transtation and gene therapy in parkinsonian monkeys. Neurosci Abs2001;27(245):243.

ubramanian T, Deogaonkar M, Brummer M, Bakay R. MRI guidaimproves accuracy of stereotaxic targeting for cell transplantationgene therapy in parkinsonian monkeys. Exp Neurol 2005;193:17

alairach J, Tournoux P. Co-planar sterotaxic atlas of the humanNew York: Thieme; 1988.

heodosopoulos PV, Marks Jr WJ, Christine C, Starr PA. Locationmovement-related cells in the human subthalamic nucleus in Pson’s disease. Mov Disord 2003;18:791–8.

oga AW, Thompson PM. Maps of the brain. Anat Rec 2001;265:37abek M, Sobstyl M, Koziara H. Bilateral deep brain stimulation

subthalamic nucleus STN in the surgical treatment of Parkindisease. Neurol Neurochir Pol 2003;37:447–55, discussion 455