This study w

Publique-Hopita

Martinot. This s

Jamila Ando

and from the ‘‘E

Corresponde

E-mail addre

Submitted O

1935-861X/09/$

doi:10.1016/j.br

Brain Stimulation (2009) 2, 123–31

www.brainstimjrnl.com

ORIGINAL RESEARCH

A triangulation-based magnetic resonanceimage-guided method for transcranialmagnetic stimulation coil positioning

Jamila Andoh, PhDa,b,c,d, Denis Riviere, PhDe, Jean-Francois Mangin, PhDa,b,c,e,Eric Artiges, PhDa,b,c,f, Yann Cointepas, PhDe, David Grevent, MDa,b,c,Marie-Laure Paillere-Martinot, MD, PhDa,b,c,g, Jean-Luc Martinot, MD, PhDa,b,c,Arnaud Cachia, PhDa,b,c

aInserm, U797 Research Unit ‘‘Neuroimaging and Psychiatry,’’ IFR49, Orsay, FrancebCEA, ‘‘Neuroimaging and Psychiatry’’ U797 Unit, Hospital Department Frederic Joliot and Neurospin, I2BM,Orsay, FrancecUniv Paris-Sud, UMR U797, Orsay; and University Paris Descartes, UMR U797, Paris; FrancedBrain and Body Centre, University of Nottingham, Nottingham, UKeImage Analysis and Structural Anatomy Group, Neurospin, IFR 49, Atomic Energy Commission, Gif/Yvette, FrancefPsychiatry Department, 91G16, Orsay Hospital, Orsay, FrancegAP-HP, Department of Adolescent Medicine, Maison de Solenn, Cochin Hospital, Paris, France

Transcranial magnetic stimulation (TMS) is currently used for cognitive studies and investigated asa treatment for psychiatric disorders. Because of the cortex variability, the coil positioning stage isdifficult and should be improved by using individual neuroimaging data. Sophisticated and expensiveneuronavigation systems have been developed to guide the coil to selected regions on the patient’smagnetic resonance images (MRI). Our objective was to develop a triangulation-based MRI-guidedmethod to position manually the TMS coil over the subject’s head, using a cortical target derived fromindividual MR data. We evaluated both the spatial accuracy and the reproducibility of the method usingfunctional MR activations of two different targets in the motor and parietal cortices. The accuracy ofthe MRI-guided method, assessed from the Euclidean distance (Dm) between the thumb motor targetand the coil position eliciting reproducible thumb motor-evoked potentials with TMS, was Dm 5 10 6

3 mm. The reproducibility of the method, evaluated across two different operators, was Dm 5 6.7 6 1.4

as funded in part by a grant from the ‘‘Programme Hospitalier de Recherche Clinique’’, ‘‘Delegation a la Recherche Clinique de l’Assistance

ux de Paris (AP-HP) (PHRC AOM 98099)’’ and by an INSERM-PROGRESS grant and an APHP/INSERM interface grant to ML Paillere-

tudy was also supported in part by a grant (APV05137LSA) from the National Agency for Research.

h was supported by a PhD grant from the ‘‘Fondation pour la Recherche Medicale.’’ Arnaud Cachia was supported by grants from INSERM

lectricite de France - EDF Foundation.’’

nce: Dr. Jean-Luc Martinot, U797 INSERM-CEA, Service Hospitalier Frederic Joliot, CEA, 4, place du General Leclerc, 91401 Orsay, France.

ss: jean-luc.martinot@cea.fr

ctober 30, 2007; revised October 29, 2008. Accepted for publication October 30, 2008.

-see front matter � 2009 Elsevier Inc. All rights reserved.

s.2008.10.002

124 J. Andoh et al

mm for the repositioning in the motor cortex and Dm 5 6.0 6 3.2 mm in the parietal cortex. This novelmethod could be used clinically to assist positioning of the TMS coil.� 2009 Elsevier Inc. All rights reserved.

Keywords transcranial magnetic stimulation; functional magnetic resonance imaging; coil posi-tioning; motor cortex

Transcranial magnetic stimulation (TMS) has becomea widely used noninvasive method to investigate cognitivefunctions, and has potential therapeutic effects for psychi-atric disorders.1,2

In many studies, the placement of the TMS coil abovethe proper site of the cortex appears to be critical. Forexample, most clinical trials for depression have targetedthe dorsolateral prefrontal cortex, Brodman’s area (BA) 9and 46, by moving forward the coil 5 cm ahead of theoptimal site for stimulation of the motor cortex.3 However,the target position was reported to vary between subjects,ranging from BA 9-BA 6, which highlights the need forindividually guided coil positioning.4

Neuronavigation systems have been developed to guidethe coil to selected regions on individual anatomic and/orfunctional magnetic resonance images (fMRI).5-8 However,these sophisticated systems are expensive for routine clinicaluse. Here, we developed a triangulation-based MRI-guidedmethod that allows users to manually position the TMScoil over a cortical target derived from functional or anatomicimages, as previously described.9 For each subject, the thumbmotor cortex location was determined with fMRI duringa thumb motor task derived from our previous studies.9,10

To determine the accuracy of the method, we used the projec-tion on the subject’s head surface of the thumb motor cortexand tested whether its stimulation produced a measurablemuscle twitch and motor-evoked potentials (MEPs) relatedto quantifiable excitability.11,12 Our objective was to estimatethe variability in distance between the scalp projection of thethumb motor cortex determined using fMRI (fMRI-target),and the coil position eliciting reproducible MEPs (MEP-target), allowing determination of the motor threshold (MT).

Methods and materials

Subjects

Ten healthy male subjects were included (24 6 4 years old)recruited from Orsay University students (France); theywere right-handed according to the Annett’s question-naire,13 (range: 90-100%; mean 96.5% 6 3.4%). Exclusioncriteria were alcoholism, drug addiction, and history ofpsychiatric or neurologic disease, as assessed by a clinicalinterview with a senior psychiatrist (J.L.M.).

All TMS experiments followed safety guidelines14,15

and approval was obtained from the local Paris-Pitie-Salpe-triere ethics committee. Each subject gave written informed

consent to participate in the study and was screened bya senior physician (J.L.M.).

MRI protocol

Task descriptionWe used an fMRI-task derived from our previous studies,showing a reliable and robust activation of the right thumbmotor cortex in each subject by using an event-relateddesign.9,10 Ten participants made thumb opposition move-ments with their left or right thumb after a word listeningtask in an fMRI scanner. Images of parameter estimateswere created for each subject for detection of right thumbmotor activation from the t-contrast (t-map) between rightand left thumb movements. The fMRI task design used wasconvenient for our study as the contrast between left and righthand motor responses subtracted the activation because oflanguage processing, isolating the motor activations.

Anatomic acquisitionA standard T1-weighted three-dimensional (3D) image wasacquired for each participant with a high-resolution headarray coil (8-channel design, Signa 1.5 T-GE) using a fastgradient echo sequence (FOV 5 24 3 18 cm, 256 3 192matrix size, 0.94 3 0.94 3 1.3 mm thick voxels). These high-resolution fSPGR T1-weighted images enabled reconstruc-tion of fine individual cortical folds.16

fMRI acquisition and analysesA gradient echo planar image (EPI) pulse sequence wasthen acquired in an interleaved sequence (20 slices, 3.75 3

3.75 3 5 mm, TR/TE 5 2000/60 milliseconds) duringa thumb motor task adapted from Andoh et al.10 Eventswere synchronized by using E-Prime software (PsychologySoftware Tools Inc, Pittsburgh, PA) with event-relatedfMRI volume acquisitions and administered in two 14-minute sessions, each including 108-randomized trials: 54right thumb motor responses and 54 left thumb motorresponses.

Functional time series data were processed using SPM2(http://fil.ion.ucl.ac.uk/spm) with MATLAB 6.5 (http://www.mathworks.com/products/matlab) and included slicetiming, spatial realignment. EPI data were smoothed witha 5-mm full width at half maximum (FWHM), adapted todetect small areas like the thumb motor cortex. For statis-tical analyses, a linear model was generated and definedby two categories of events: right and left motor responses.Images of parameter estimates for the contrast of interest

A new method for TMS coil positioning 125

were created for each subject and right thumb motor activationwas detected from the t-contrast (t-map) between right and leftthumb movements (voxelwise threshold, P , .0001 FWE-corrected; cluster size threshold P , .00001 threshold, 10contiguous voxels). The MRI-guided method was then usedto individually position the TMS coil over the right thumbmotor target on the subject’s head surface.

MRI-guided method

The MRI-guided method is embedded as a plug-in inBrainvisa, a software bundle released by a group of publicresearch institutes (CEA, ‘‘French Atomic Energy Commis-sion’’; INSERM, ‘‘National Institute of Health and MedicalResearch’’; and CNRS, ‘‘French National Centre for Scien-tific Research’’). Brainvisa is freely available for download(http://brainvisa.info/) and is shipped as a single completepackage, which contains many image processing tools, aim-ing at anatomic and functional brain processing. These toolswere used in this study to provide realistic and accurate 3Drepresentations of the head and the brain, and allow foreasy localization on the cortex and thus definition ofanatomic and functional landmarks in each subject.16

The MRI-guided method used individual MRIs todetermine the TMS coil position onto the head surface(software procedure available at ftp://ftp.cea.fr/pub/dsv/anatomist/documents/TMS/Andoh_TMS-positioning.pdf).

In the following section, we present the MRI protocolproviding anatomic and fMRIs acquired during a thumbmotor task, used to individually define the right thumbmotor target. We describe afterward the procedure of theMRI-guided method to position the TMS coil over the rightthumb motor target on the subject’s head surface.

Procedure of the MRI-guided method

The procedure of the MRI-guided method required threesteps:

Figure 1 Cortical target projection onto head surface. A, Locally sphedetermine, as all the mesh nodes (yellow nodes) are equidistant to the creconstruction. The closest scalp mesh node (yellow node) may not coations, the barycenter of the n (n 5 10) closest nodes provides a robuthe neighbor mesh nodes.

1. The first step was the segmentation of the head surface.MRI intensities were preliminary thresholded from theintensity histogram.17,18 Then, a morphologic closing19

was applied to the binary result to fill up cavities and holesinside the head. Finally, a smooth mesh of the resultingobject was computed using a marching-cube algorithm.20

2. The second step was the projection of the cortical targetonto the 3D reconstruction of the head surface. Thecortical target (eg, local maximum of activation cluster)was automatically projected onto the 3D head meshsurface by using a classical 3D Euclidean distance(Figure 1A). As inaccuracies in head surface reconstruc-tion may induce local geometric artifacts, and confoundtarget projection onto the head surface, we did not usethe closest scalp mesh node but the center of gravity ofthe (n 5 10) closest nodes to obtain a robust projectionof the cortical target onto the head surface (Figure 1B).

3. The third step was the computation of the geodesicdistances (ie, tangent to the head surface) between this pro-jected target on the surface and landmarks on the subject’shead (nose bridge, tragus of both ears, Figure 2). Thesethree distances (mm) were then used to triangulate and toposition the TMS coil manually over the subject’s head.This position was marked physically on the cap of eachsubject (Figure 3). The triangulation was performed manu-ally by the same investigator (J.A.) for all subjects, usingthree tied inextensible electric threads (ref: FILEctriquesCAbles, v/Ref: 7213120270, diameter 0.22 mm2), whichlength corresponded to the three geodesic distances. Afourth thread was used to draw and extent the other threadsover the subject’s head landmarks (Figure 4).

The duration of the localization calculations (head surfacereconstruction, computation of the geodesic distancesbetween the target and head landmarks) lasted about 20minutes per subject. The total duration for the experiment(localization calculations and manual placement) lastedabout 30 minutes per subject.

rical head surface. The closest scalp mesh node may be difficult toortical target (blue). B, Locally geometric artifacts in head surfacerrespond to the optimal position for projected target. In both situ-st definition of the projected target (red dot). Grey dots represent

Figure 2 Projection of the functional cortical target (right thumb motor area). A, Individual statistical parametrical map for the functionalmagnetic resonance image (fMRI) subtraction ‘‘right minus left thumb’’ providing the thumb motor target. The cortex cluster with the highestt-value was clearly visible in the subject displayed and corresponded to Brodman’s area BA4 (peak voxel at: 232, 225, 57; Z . 6.65; extent:1536 mm3) in the MNI space. B, The projected peak voxel on the subject’s three-dimensional head surface (Figure 1) was used to estimate thethree geodesic distances to head landmarks. C, These distances allowed operators to guide the coil over the right thumb motor area.

126 J. Andoh et al

Validation of the MRI-guided method

We assessed the accuracy of the MRI-guided method usinga standard approach based on a right thumb motor cortextarget obtained from fMRI.8,21-24 The coil position obtainedby the MRI-guided method was used as a gross estimationof the actual MEP-target. A refinement of the coil positionwas obtained iteratively with a heuristic procedure, withdistance steps ranging from 2 cm to 1 mm, until objectivedetection of the MT, namely, the lowest stimulation inten-sity for evoking at least five MEPs in 10 stimulations ofat least 50 mV recorded by a surface electromyogram(EMG) (Keypoint Portable, Medtronic Ltd, Hertfordshire,UK) from the relaxed right abductor pollicis brevis25-27

was reached. We then compared the coil position over theright thumb motor target obtained from the MRI-guidedmethod (fMRI-target) with the coil position obtained withthe usual MT determination method, eliciting reproduciblethumbs MEPs with stimulation (MEP-target). To quantifythe coil position for validation purpose, the coordinates ofthe coil positions over the fMRI-target and over theMEP-target were obtained for each subject, with the opticaltracking system included in Brainsight Frameless (TheMagstim Company Ltd, Carmarthenshire, Wales, UK),following a standardized calibration procedure that consistsin a registration between the physical space (subject’s head)and the image space (individual MRI). The targets’ loca-tions were determined for each subject using individualcoordinates without spatial normalization.

Measures of MEPs in thumb abductor during TMS

The target position eliciting thumb MEPs in response tomagnetic stimulation was defined according to MT criteria(described previously) and its coordinates were recordedonline with Brainsight system (The Magstim Company

Ltd). Magnetic stimulation was applied with a MagProStimulator (http://www.medtronic.com) using a figure-8coil. The spot of the figure-8 coil (ie, the junction betweenthe two circular coils) was calibrated using the Brainsightsystem (The Magstim Company Ltd). Stimulus intensitieswere increased in steps of 1% stimulator output startingfrom 50% of the maximal stimulator output.

Accuracy of the MRI-guided method

The accuracy of the MRI-guided method was characterizedby the mean Euclidean distance (Dm) between the theoret-ical coil position over the right thumb motor target obtainedfrom the MRI-guided method (fMRI-target) and the coilposition obtained after evoking thumb MEPs with stimula-tion (MEP-target).

To obtain descriptive statistics overcoming the globalvariability in size and shape between the subjects’ heads, thecoordinates of fMRI-target and MEP-target were linearlytransformed into a common reference (MNI space).28 Thecorresponding 3D MNI coordinates of the MRI-guidedmethod (xfMRI, yfMRI, zfMRI) and the position elicitingMEPs (xMEP, yMEP, zMEP) refer, respectively, to left-right,posterior-anterior, and bottom-up directions. The computa-tion of the 3D Euclidean distance in MNI space betweenfMRI and MEP targets defined as D5 [(xfMRI – xMEP)2 1

(yfMRI – yMEP)2 1 (zfMRI – zMEP) 2]1/2 5 [Dx2 1 Dy

2 1

Dz2]1/2, was used to quantify the spatial accuracy of the

MRI-guided method.22-24

Reproducibility of the MRI-guided method

A complementary experiment was carried out to evaluatethe between-operator and between-target reproducibility ofthe method. Five healthy subjects were included (22 6 1year old; 4 men, 1 woman) and recruited from students at

Figure 3 Positioning of the functional magnetic resonanceimage (fMRI)-target over the subject’s head. The triangulationwas performed manually using three tied threads (dark grey onthe figure) whose length corresponded to the three geodesicdistances (provided by Brainvisa) between the head landmarks(indicated by black arrows) and the fMRI-target (right thumbmotor cortex). The target position was marked physically on thecap of each subject.

Figure 4 Representation of the four threads used for the posi-tioning of the functional magnetic resonance image (fMRI)-target.Three tied inextensible electric threads (ref: FILEctriques CAbles,v/Ref: 7213120270, diameter 0.22 mm2) were used for the trian-gulation. The fourth thread was used to draw and extent the otherthreads over the subject’s head landmarks.

A new method for TMS coil positioning 127

Ecole Polytechnique, Palaiseau (France). One was left-handed according to the Annett’s questionnaire.13

The method reproducibility was evaluated using twodifferent targets: (1) the motor cortex of the right thumb,which was functionally defined using fMRI during a thumbmotor task and (2) a target located more posterior in the rightparietal cortex, functionally defined using a mental rotationtask similar to the one used by Cooper and Shepard.29 Twooperators used the MRI-guided method blindly and indepen-dently to position consecutively the motor cortex target andthe right parietal target over the cap of each subject’s head.

To localize the target in the right parietal cortex, subjectswere asked to participate in a mental rotation task. Stimuliwere all asymmetrical alphabetic characters: two uppercaseletters (L, G) and one lowercase (a). The three letters (L, G,a) to be judged normal or mirror image were rotated at fourdifferent angular orientations, ranging in 60-degree incre-ments from 0-180 degrees clockwise from the upright. Athird ‘‘resting’’ or ‘‘fixation’’ condition was added in whichsubjects were asked to maintain their gaze on a smallcentrally located crosshair.

These stimuli were presented in three separate sets ofblocks. Within each block, a letter was presented three times,once each in normal or mirror direction at four differentorientations. Thus, there were 24 possible letters (5 3 32 3

4). Each bloc included 32 trials, including 24 letters and 8crosshair stimuli. A total experiment consisted of thepresentation of three blocs (one for each letter), resulting ina total of 96 stimuli (32 33) presented during 10 minutes and19 seconds.

Functional images were acquired in each plane withinterleaved slice order (TR/TE 5 2400/30, FOV 5 24 cm,64 3 64 matrix, voxel size 3.75 3 3.75 3 3 mm). A total of258 images were acquired for each subject using a T2*-weighted gradient echo sequence. Images of parameterestimates for conditions of interest were created for eachsubject for detection of right parietal changes in task-related activity resulting from the t-contrast betweennormal and mirror images.

Results

The statistical parametrical map (SPM) reflecting rightthumb movement activity detected a clear cortical activa-tion for each subject in the precentral gyrus in the region ofthe hand knob, providing the thumb motor target (Figure 2).In all subjects, the MRI-guided method allowed operatorsto induce MEPs while stimulating the thumb motor target.In two subjects, the MT was found while stimulating thistarget, with an MEP amplitude . 50 mV in 6 of 10 trialsfor subject 2, and in 8 of 10 trials for subject 6. In the othersubjects, the coil was repositioned to find the MT (Table 1).

The mean Euclidean distance (m 6 SD) between thefMRI and the MEP targets was Dm 5 10.1 6 2.9 mm,ranging between 5.2 and 14.0 mm (Table 1). The compo-nents of this distance in the x, y, and z-axes were Dx5

4.0 6 2.2 mm (range 0.6-7.7mm), Dy 5 5.0 6 2.5 mm(range 1.3-8.0 mm) and Dz 5 7.1 6 3.0 mm (range 2.8-11.1 mm).

Thumb MTs ranged between 50% and 65% of stimulatoroutput (mean 58% 6 5%).

The results of the repositioning of the motor cortex targetacross two different operators showed a mean Euclideandistance of 6.7 6 1.4 mm, ranging between 5.4 and 8.3 mm.In all subjects, the MRI-guided method allowed operators to

Table 1 Euclidean distances (in individual and MNI spaces) between the coil position over the right thumb area given by the MRI-guidedmethod (fMRI target), and the coil position eliciting reproducible thumb motor-evoked potentials in response to transcranial magneticstimulation pulses (MEP target)

Individual space MNI space

Individual coordinates (mm) MNI coordinates (mm)

Subjects Dx Dy DzEuclideandistance D (mm) Dx Dy Dz

Euclideandistance D (mm)

Motorthreshold (% SO)

S1 7.5 5.0 5.5 10.6 7.0 4.9 5.4 10.1 50S2 1.9 6.0 2.5 6.7 1.8 5.7 3.2 6.7 58S3 3.9 4.7 7.9 10.0 3.7 5.1 10.0 11.9 58S4 4.2 8.9 6.5 11.9 4.0 8.1 8.8 12.6 51S5 7.0 4.8 13.9 16.4 3.9 7.8 9.1 12.6 60S6 3.1 1.0 3.0 4.5 3.2 1.3 3.9 5.2 65S7 8.3 1.2 2.1 8.6 7.7 1.8 2.8 8.4 56S8 5.7 7.1 8.5 12.5 5.4 6.7 11.1 14.0 61S9 0.6 7.3 7.1 10.2 0.6 6.9 9.3 11.6 56S10 3.4 2.1 5.5 6.8 3.2 1.8 7.3 8.2 63

Euclidean distances in a common space (I, MNI space) were required to perform descriptive statistics on the whole group independently from possible global

differences in size and shape between the subjects’ heads. The motor threshold for each subject is provided in percentage of stimulator output (% SO).

128 J. Andoh et al

induce a motor response. The results of the repositioning overthe right parietal target across operators was determined onfour subjects because parietal activation was not detected inone subject, and showed a mean Euclidean distance of 6.0 6

3.2 mm, ranging from 1.4-8.7 mm (Table 2).

Discussion

We developed a novel MRI-guided method using individualbrain MRIs to position the TMS coil reliably on the subject’shead. The findings show that the accuracy of the method wasabout 1 cm (Dm 5 10.1 6 2.9 mm), with a nonsystematicdeviation. Indeed, the difference between the localisationof the fMRI-target was either anterior or posterior from theMEP-target. This accuracy is analogous to that of other guid-ance systems using neuronavigated figure-8 coils and fMRI-defined cortex targets in human volunteers: Dm 5 9.8 6 7.5mm,24 Dm 5 9.5 6 3.2 mm.30 This accuracy is also consistentwith the focality of the TMS coil (eg, theoretically 2-3 cm2 at

Table 2 MNI coordinates and mean Euclidean distance (in MNI space)different targets

Motor cortex target Right

MNI coordinates (mm) MNI c

Subjects Dx Dy DzEuclideandistance D (mm) Dx

S1 3.1 4.5 5.9 8.1 0.1S2 2 4.5 2.3 5.4 1.2S3 2.6 3.2 4.9 6.4 2.9S4 2.6 4 2.9 5.6 2.5S5 0.2 6.2 5.4 8.3

Motor cortex target: thumb motor area in the left hemisphere and right parieta

provided in percentage of stimulator output (% SO).

110% intensity of the MT).31 The differences between theintensity of stimulation used by Thielscher et al31 and thatused here (ie, 58% 6 5% of stimulator output) may explainthat for many subjects (n 5 8), the site of the fMRI methodstimulation did not lie within the area in which the MEPmethod elicited a response.

Other approaches have been used to assess the accuracy ofthe TMS coil positioning. Lotze et al24 used vitamin E capsulesas markers to indicate the location of MEP sites, and found thatthe mean distance from the projection of the fMRI activationmaxima and vitamin E capsules over TMS highest MEP was18.8 mm. Herwig et al23 found a mean distance of 9.8 mmusing a neuronavigation system to measure distances betweena cortical area activated by a motor task in fMRI and the areaof TMS MEP using a 5 3 5 mm spaced grid.

In comparison with these approaches, the currentmethod may have some advantages because it is relativelyinexpensive and does not require any additional experi-mental setting. In addition, the MRI-guided method couldbe particularly useful for therapeutic protocols in patients

of the target repositioning across two different operators and two

parietal target

oordinates (mm)

Dy DzEuclideandistance D (mm)

Motorthreshold (% SO)

6.5 5.7 8.7 600.7 0.1 1.4 451.9 4.8 6 607 2.6 7.9 55

50

l target: in the right hemisphere. The motor threshold for each subject is

A new method for TMS coil positioning 129

included in multicenter trials, as the three distances can beused offline to position the TMS coil. Moreover, it onlyneeds Brainvisa, a software freely available online and isrelatively user-friendly (a maximum of 3 hours is sufficientto get someone to implement the method).

The reproducibility of the MRI-guided method was6.7 6 1.4 mm over the thumb motor cortex and 6.0 6

3.2 mm over the right parietal target, which is comparableto its spatial accuracy (around 1 cm). Few studies haveevaluated the spatial reproducibility of their method acrossdifferent sessions. Schonfeldt-Lecuona et al32 determinedthe intersession variability of an optically tracked framelessstereotaxic system by measuring the mean Euclideandistance between the coordinates of head landmarks (ie,nose, ears, and eyes). They found an average deviation of2.5 mm between 2 consecutive days. Chronicle et al33

used a TMS coil holder and demonstrated an average devi-ation of 2 mm between two consecutive TMS sessionsapplied over the motor cortex. Stereotaxic positioningseems to be more stable and reliable for TMS positioningthan manual positioning. However, the factor of a possiblemanual imprecision during referencing procedures, and/orcoil placement should be considered regarding the generallimitations of positioning accuracy.32

Limitations

The results reported here should be considered in light ofsome methodologic and technical limitations caused byboth the MRI-guided method used to position the coil, andby the procedure used to validate this method.

First, the current sample size is relatively small,although the number of subjects was analogous to that ofother studies.8,22,24

Second, the current MRI-guided method is not a neuro-navigation system and therefore cannot provide onlinemonitoring such as a real-time control of the coil angle.However, it is noteworthy that the influence of the coil angleremains debated.34,35 Hence, we placed the coil tangentiallyto the scalp with the handle pointing backwards and rotatedfrom the midline by 45 degrees as in previous studies.36,37

Then, the positioning of the MEP target was determinediteratively with a heuristic procedure, which could be not asreliable as studies that used a grid-point involving therepresentational map of the motor cortex.8,24 However, ourdefinition of the MEP-target is objective as it is based on anoperator-independent criterion: the MT.25-27 The position ofthe MEP-target was not determined blindly by the investigatorbecause the fMRI-target was determined before the MEP-target. However, this iterative approach is unlikely to havebiased the estimation of the MEP-target coordinates, as thistarget was defined from the objective criterion of the MT.Indeed, the position of the MT site does not need to be definedblind to the position obtained by triangulation from fMRIbecause the MTis defined from an operator-independent crite-rion (MEPs). MT is an objective measure, commonly used

and referenced,25-27 although it might be less accurate thanthe MEP center of gravity used by previous authors.8,24

In addition, TMS and fMRI give related but differentinformation as the neuron population causing the blood-oxygen-level dependent (BOLD) signal and the neuronpopulation activated by TMS may not be identical. Indeed,TMS is activating all excitable circuits within the volumereached by the induced current, whereas fMRI investigateschanges in blood flow in response to motor tasks.21,24,38 Forinstance, Rutten et al39 showed a discrepancy (approxi-mately 5 mm) between fMRI-based localization of brainfunction and the location determined from direct corticalstimulation during surgical procedures.

Moreover, despite the results are interpreted as evidencingdistance differences between the two methods, is it possiblethat results may not only be reflecting methodologic/technical aspects of the procedures but also the fact that thethumb responses during the fMRI task and the motor-evokedresponses induced by TMS are not directly comparable.Indeed, the responses probably do not involve the exact samegroup of muscles, and it is possible that they do not imply thesame exact spatial representation in the primary motorcortex. The fact that we were using an ‘‘externally induced’’response (MEP) with one method and a voluntary muscularresponse with the other raises the possibility that the nature ofthe responses could be a confounding variable when trying tocompare directly the two methods. For instance, Denslowet al40 examined the differences between function-guidedand image-guided TMS. Authors observed a high variabilityin coil functional regions and placement relative to fMRIsignal location that may indicate that widely spaced andpossibly disjoint cortical sites result in fMRI signal at a singlesite. In an interleaved TMS/fMRI study, Denslow et al41

elucidated whole brain activation patterns from 1-Hz TMSover left primary motor cortex. The authors showed somedifferences between time courses of BOLD intensity duringTMS circuit activation and volitional circuit activation.

In line with these findings, Niyazov et al35 comparedfMRI activations during executed and imagined movementsof the index finger with TMS-induced MEPs. Resultsshowed a discrepancy between executed movements of1 cm posterior to the TMS, whereas the imaged movementclosely agreed with the TMS-induced MEPs. The authorssuggested that the discrepancy between fMRI and TMS-induced MEPs might be due to involvement of the somato-sensory component in the executive movement.

Finally, the mechanisms of the neuronal depolarizationjust beneath the TMS coil remains unclear as the depolarizedneurons for physical reasons of the electromagnetic fieldcould not be directly beneath the center of the coil.42,43

Conclusion

In summary, the triangulation-based MRI-guided methodpresented herein provides a reliable and inexpensive way to

130 J. Andoh et al

position the TMS coil that may be used in case ofunavailability of online neuronavigation, for instance, ina clinical setting. This MRI-guided method is versatile andcan use cortical landmarks from either functional oranatomic image targets. Also, it has a 3D graphic interfaceand could be used for research on physiology or neuropsy-chiatric conditions.

The authors thank the subjects who participated in thestudy, especially the students from Polytechnique for theirdynamism and motivation. Pr. Andre Syrota is acknowl-edged for his support.

References

1. Haraldsson HM, Ferrarelli F, Kalin NH, et al. Transcranial magnetic

stimulation in the investigation and treatment of schizophrenia:

a review. Schizophr Res 2004;71:1-16.

2. Rossini PM, Rossi S. Transcranial magnetic stimulation: diagnostic,

therapeutic, and research potential. Neurology 2007;68:484-488.

3. George MS, Wassermann EM, Williams WA, et al. Daily repetitive

transcranial magnetic stimulation (rTMS) improves mood in depres-

sion. Neuroreport 1995;6(14):1853-1856.

4. Herwig U, Padberg F, Unger J, et al. Transcranial magnetic stimula-

tion in therapy studies: examination of the reliability of ‘‘standard’’

coil positioning by neuronavigation. Biol Psychiatry 2001;50:58-61.

5. Herwig U, Schonfeldt-Lecuona C, Wunderlich AP, et al. The naviga-

tion of transcranial magnetic stimulation. Psychiatry Res 2001b;108:

123-131.

6. Krings T, Buchbinder BR, Butler WE, et al. Stereotactic transcranial

magnetic stimulation: correlation with direct electrical cortical stimu-

lation. Neurosurgery 1997;41:1319-1325.

7. Paus T, Wolforth M. Transcranial magnetic stimulation during PET:

reaching and verifying the target site. Hum Brain Mapp 1998;6:

399-402.

8. Neggers SF, Langerak TR, Schutter DJ, et al. A stereotactic method

for image-guided transcranial magnetic stimulation validated with

fMRI and motor-evoked potentials. Neuroimage 2004;21:1805-1817.

9. Andoh J, Artiges E, Pallier C, et al. Modulation of language areas with

functional MR image-guided magnetic stimulation. Neuroimage 2006;

29:619-627.

10. Andoh J, Artiges E, Pallier C, et al. Priming frequencies of transcra-

nial magnetic stimulation over Wernicke’s area modulate word detec-

tion. Cerebral Cortex 2008;18(1):210-216.

11. Rothwell JC. Techniques and mechanisms of action of transcranial

magnetic stimulation of the human motor cortex. J Neurosci Methods

1997;74:113-122.

12. Boroojerdi B, Foltys H, Krings T, et al. Localization of the motor hand

area using transcranial magnetic stimulation and functional magnetic

resonance imaging. Clin Neurophysiol 1999;110:699-704.

13. Annett M. The binomial distribution of right, mixed and left handed-

ness. Q J Exp Psychol 1967;19:327-333.

14. Chen R, Gerloff C, Classen J, et al. Safety of different inter-train inter-

vals for repetitive transcranial magnetic stimulation and recommenda-

tions for safe ranges of stimulation parameters. Electroencephalogr

Clin Neurophysiol 1997;105:415-421.

15. Wassermann EM. Risk and safety of repetitive transcranial magnetic

stimulation: report and suggested guidelines from the International Work-

shop on the Safety of Repetitive Transcranial Magnetic Stimulation, June

5-7, 1996. Electroencephalogr Clin Neurophysiol 1998;108:1-16.

16. Riviere D, Mangin JF, Papadopoulos-Orfanos D, et al. Automatic

recognition of cortical sulci of the human brain using a congregation

of neural networks. Med Image Anal 2002;6:77-92.

17. Mangin JF, Riviere D, Cachia A, et al. A framework to study the

cortical folding patterns. Neuroimage 2004;23:129-138.

18. Mangin JF, Coulon O, Frouin V. Robust brain segmentation using

histogram scale-space analysis and mathematical morphology. In

Proc. 1st MICCAI, LNCS-1496, MIT, Boston; 2005. p. 1230-1241.

19. Serra J . Image analysis and mathematical morphology. Vol. I. London:

Academic Press; 1982. p. 600.

20. Lorensen WE, Cline HE. Marching cubes: a high resolution 3D

surface construction algorithm. computer graphics. Proc SIGGRAPH

1987;21(4):163-169.

21. Bastings EP, Gage HD, Greenberg JP, et al. Co-registration of cortical

magnetic stimulation and functional magnetic resonance imaging.

Neuroreport 1998;9:1941-1946.

22. Gugino LD, Romero JR, Aglio L, et al. Transcranial magnetic stimu-

lation coregistered with MRI: a comparison of a guided versus blind

stimulation technique and its effect on evoked compound muscle

action potentials. Clin Neurophysiol 2001;112:1781-1792.

23. Herwig U, Kolbel K, Wunderlich AP, et al. Spatial congruence of neu-

ronavigated transcranial magnetic stimulation and functional neuroi-

maging. Clin Neurophysiol 2002;113:462-468.

24. Lotze M, Kaethner RJ, Erb M, et al. Comparison of representational

maps using functional magnetic resonance imaging and transcranial

magnetic stimulation. Clin Neurophysiol 2003;114:306-312.

25. Pascual-Leone A, Houser CM, Reese K, et al. Safety of rapid-rate

transcranial magnetic stimulation in normal volunteers. Electroenceph

Clin Neurophysiol 1993;89:120-130.

26. Rossini PM, Barker AT, Berardelli A, et al. Non-invasive electrical

and magnetic stimulation of the brain, spinal cord and roots: basic

principles and procedures for routine clinical application. Report of

an IFCN committee. Electroencephalogr Clin Neurophysiol 1994;

91:79-92.

27. Rothwell JC, Hallett M, Berardelli A, et al. Magnetic stimulation:

motor evoked potentials. Electroencephalogr Clin Neurophysiol Suppl

1999;52:97-103.

28. Friston KJ, Holmes AP, Worsley KJ, et al. Statistical parametric maps

in functional imaging: a general linear approach. Hum Brain Mapp

1995;189-210.

29. Cooper AN, Shepard RN. The time required to prepare for a rotated

stimulus. Mem Cognit 1973;1:246-250.

30. Lancaster JL, Narayana S, Wenzel D, et al. Evaluation of an image-

guided, robotically positioned transcranial magnetic stimulation

system. Hum Brain Mapp 2004;22:329-340.

31. Thielscher A, Kammer T. Electric field properties of two commercial

figure-8 coils in TMS: calculation of focality and efficiency. Clin Neu-

rophysiol 2004;115:1697-1708.

32. Schonfeldt-Lecuona C, Thielscher A, Freudenmann RW, et al. Accu-

racy of stereotaxic positioning of transcranial magnetic stimulation.

Brain Topogr 2005;17(4):253-259.

33. Chronicle EP, Pearson AJ, Matthews C. Development and positioning

reliability of a TMS coil holder for headache research. Headache

2005;45(1):37-41.

34. Dubach P, Guggisberg AG, Rosler KM, et al. Significance of coil

orientation for motor evoked potentials from nasalis muscle elicited

by transcranial magnetic stimulation. Clin Neurophysiol 2004;115:

862-870.

35. Niyazov DM, Butler AJ, Kadah YM, et al. Functional magnetic reso-

nance imaging and transcranial magnetic stimulation: effects of motor

imagery, movement and coil orientation. Clin Neurophysiol 2005;116:

1601-1610.

36. Gerschlager W, Siebner HR, Rothwell JC. Decreased corticospinal

excitability after subthreshold 1 Hz rTMS over lateral premotor

cortex. Neurology 2001;57:449-455.

37. Hoffman RE, Hawkins KA, Gueorguieva R, et al. Transcranial

magnetic stimulation of left temporoparietal cortex and medication-

resistant auditory hallucinations. Arch Gen Psychiatry 2003;60:

49-56.

A new method for TMS coil positioning 131

38. Attwell D, Iadecola C. The neural basis of functional brain imaging

signals. Trends Neurosci 2002;25:621-625.

39. Rutten GJ, Ramsey NF, van Rijen PC, et al. Development of a functional

magnetic resonance imaging protocol for intraoperative localization of

critical temporoparietal language areas. Ann Neurol 2002;51:350-360.

40. Denslow S, Bohning DE, Bohning PA, et al. An increased precision

comparison of TMS-induced motor cortex BOLD fMRI response for

image-guided versus function-guided coil placement. Cogn Behav

Neurol 2005;18:119-126.

41. Denslow S, Lomarev M, George MS, et al. Cortical and subcortical

brain effects of transcranial magnetic stimulation (TMS)-induced

movement: an interleaved TMS/functional magnetic resonance

imaging study. Biol Psychiatry 2005;57(7):752-760.

42. Bestmann S, Baudewig J, Siebner HR, et al. Functional MRI of the

immediate impact of transcranial magnetic stimulation on cortical

and subcortical motor circuits. Eur J Neurosci 2004;19(7):

1950-1962.

43. Hada Y, Abo M, Kaminaga T, et al. Detection of cerebral blood flow

changes during repetitive transcranial magnetic stimulation by

recording hemoglobin in the brain cortex, just beneath the stimulation

coil, with near-infrared spectroscopy. Neuroimage 2006;32(3):

1226-1230.