NeuroImage 55 (2011) 590–596

Contents lists available at ScienceDirect

NeuroImage

j ourna l homepage: www.e lsev ie r.com/ locate /yn img

Transcranial direct current stimulation over the primary motor cortex during fMRI

Andrea Antal a,⁎,1, Rafael Polania a,1, Carsten Schmidt-Samoa b, Peter Dechent b,1, Walter Paulus a,1

a Department of Clinical Neurophysiology, Georg-August University of Göttingen, 37075 Göttingen, Germanyb MR-Research in Neurology and Psychiatry, Georg-August University of Göttingen, 37075 Göttingen, Germany

⁎ Corresponding author. Department of Clinical NeUniversity of Göttingen, 37075 Göttingen, Robert Koch551 398126.

E-mail address: Aantal@gwdg.de (A. Antal).1 These authors contributed equally.

1053-8119/$ – see front matter © 2010 Elsevier Inc. Aldoi:10.1016/j.neuroimage.2010.11.085

a b s t r a c t

a r t i c l e i n f o

Article history:Received 6 October 2010Revised 8 November 2010Accepted 10 November 2010Available online 4 January 2011

Keywords:fMRItDCSMotor cortexHuman brainInhibitionFacilitation

Measurements of motor evoked potentials (MEPs) have shown that anodal and cathodal transcranial directcurrent stimulations (tDCS) have facilitatory or inhibitory effects on corticospinal excitability in thestimulated area of the primary motor cortex (M1). Here, we investigated the online effects of short periods ofanodal and cathodal tDCS on human brain activity of healthy subjects and associated hemodynamics byconcurrent blood-oxygenation-level-dependent (BOLD) functional magnetic resonance imaging (fMRI) at 3 T.Using a block design, 20 s periods of tDCS at 1 mA intensity over the left M1 altered with 20 s periods withouttDCS. In different fMRI runs, the effect of anodal or cathodal tDCS was assessed at rest or during finger tapping.A control experiment was also performed, in which the electrodes were placed over the left and right occipito-temporo-parietal junction. Neither anodal nor cathodal tDCS over the M1 for 20 s stimulation durationinduced a detectable BOLD signal change. However, in comparison to a voluntary finger tapping task withoutstimulation, anodal tDCS during finger tapping resulted in a decrease in the BOLD response in thesupplementary motor area (SMA). Cathodal stimulation did not result in significant change in BOLD responsein the SMA, however, a tendency toward decreased activity could be seen. In the control experiment neithercathodal nor anodal stimulation resulted in a significant change of BOLD signal during finger tapping in anybrain area including SMA, PM, and M1. These findings demonstrate that the well-known polarity-dependentshifts in corticospinal excitability that have previously been demonstrated using measurements of MEPs afterM1 stimulation are not paralleled by analogous changes in regional BOLD signal. This difference implies thatthe BOLD signal and measurements of MEPs probe diverse physiological mechanisms. The MEP amplitudereflects changes in transsynaptic excitability of large pyramidal neurons while the BOLD signal is a measure ofnet synaptic activity of all cortical neurons.

urophysiology, Georg-AugustStraße 40, Germany. Fax: +49

l rights reserved.

© 2010 Elsevier Inc. All rights reserved.

Introduction

Modulation of cortical excitability can be achieved by usingexternal stimulation methods such as transcranial direct currentstimulation (tDCS), also known as brain polarization. An earlyexample of the clinical application of a brain polarization techniquewas done in 1802 shortly after invention of the voltaic pile (Hellwag,1802). Over the centuries the method was tried several times andabandoned mainly due to lack of a sufficiently suitable evaluationmethod. This changed as soon as transcranial magnetic stimulation(TMS) was used for quantification of acute effects (Priori et al., 1998)and plastic aftereffects (Nitsche and Paulus, 2000). Anodal stimulationof the primary motor cortex (M1) increased the amplitude of motorevoked potentials (MEPs) while cathodal stimulation decreased them.Although MEPs are the most robust evaluation method, perceptual

effects of tDCS applied over the visual areas were also found to be inaccordance with its physiological effect and mirrored those after-effects achieved in the M1 (for a recent review see: Antal and Paulus,2008).

tDCS is now a well-established non-invasive, painless techniquefor interventional use in research with potential therapeutic use inneurorehabilitation, chronic pain, focal epilepsy, and neuropsychiatricdisorders (Webster et al., 2006; Fregni et al., 2006; Liebetanz et al.,2006; overview in Nitsche et al., 2008). tDCS induces membranepotential shifts dependent on stimulation strength, cortical layer andspatial orientation of stimulated neurons (Radman et al., 2009). Givensufficient stimulation duration the effect of stimulation can outlast theduration for several hours (Bindman et al., 1964; Nitsche and Paulus,2001).

Neuroimaging techniques have the advantage of measuringcorrelates of neuronal activity both under the stimulating electrodesand also in remote brain regions during electrical stimulation, andwas first implemented for tDCS studies using positron emissiontomography (PET) (Lang et al., 2005) however, at the expense ofradiation exposure. Combining blood oxygenation level dependent(BOLD) functional magnetic resonance imaging (fMRI) with

Fig. 1. Experimental setup for concurrent tDCS and fMRI. The DC stimulator is placedoutside the scanner room and connected to a filter box placed very close to a radiofrequency filter tube through the wall of scanner room. Via a long cable the connectionto a second filter box inside the scanner room placed on the patient table is achieved,which then connects to the electrodes attached to the head of the subject.

Fig. 2. Three-dimensional surface reconstruction (top) and axial cross-sections(bottom) from the T1-weighted anatomical dataset of a single subject. The electrodesover the left-hemispheric M1 hand area (arrowheads) and right-hemispheric orbita(arrows) are indicated. The cable running from the electrode over M1 across thesubject's head is marked by asterisks.

591A. Antal et al. / NeuroImage 55 (2011) 590–596

concurrent tDCS allows for a non-invasive detailed examination oftDCS-induced effects throughout the brain. According to our knowl-edge, there is only one study published using concurrent fMRI andtDCS (Kwon et al., 2008). In this study anodal tDCS was applied to thescalp over the precentral knob of the left M1 using 4×21 s stimulationphases (resting-tDCS-tDCS-tDCS-tDCS). No cortical activation wasdetected in any of the stimulation phases except the fourth tDCSphase. Activation was found under the electrode but also in the leftsupplementary motor cortex and the right posterior parietal cortex.Here, cathodal stimulation was not applied. In the present study,therefore, we first addressed the question as to whether anodal andcathodal tDCS result in BOLD fMRI signal changes during a restcondition. Secondly, we examined the effects of tDCS on the brainnetwork activated by a voluntary finger tapping task.

Materials and methods

Subjects

The study involved altogether 20 healthy volunteers (11 women;mean age, 25±6 years; age range, 21–32 years). 13 subjectsparticipated in the main experiment and 13 in the control session.Subjects were informed about all aspects of the experiments and allgave informed consent. Twelve of the subjects (from 20) were naivewith regard to tDCS and all of the subjects were naive with regard tothe purpose and aims of the study. None of the subjects suffered fromany neurological or psychological disorders, had metallic implants/implanted electric devices, or took anymedication regularly, and noneof them took any medication in the 2 weeks before their participationin any of the experiments. All subjectswere right handed, according tothe Edinburgh handedness inventory (Oldfield, 1971). We conformedto the Declaration of Helsinki, and the experimental protocol wasapproved by the Ethics Committee of the University of Goettingen.

Transcranial direct current stimulation (tDCS)

Direct current was provided via a pair of square rubber electrodes(7×5 cm), manufactured to be compatible with the MR-scannerenvironment. The electrodes were equipped with 5.6 kΩ resistors ineach wire to avoid sudden temperature increases due to inductionvoltages from radio frequency pulses. They were connected to aspecially developed battery-driven stimulator (NeuroConn GmbH,Ilmenau, Germany) outside the magnet room via a cable runningthrough a radio frequency filter tube in the cabin wall (Fig. 1). Twofilter boxes were placed between the stimulator and the electrodes.The characteristic bandwidth of the filters on the DC current pathwas chosen to have an approximate attenuation of 60 dB within afrequency range of 20–200 MHz to suppress the radio frequencyimpulse energy.

In order to properly position the electrodes over the M1 of thesubjects' head, the representational field of the right hand wasdetermined using suprathreshold TMS pulses. Before subjects enteredthe MR scanner, the electrodes were placed atop the respective left-hemispheric M1 hand area and above the contralateral right orbitausing conventional electrode gel (Fig. 2). For cathodal tDCS, thecathode was placed above the M1, for anodal tDCS the direction of theelectric flux was reversed. In the control experiment we testedwhether the electrical stimulation could result in any non-specificeffect due to e.g. increased attention. In order to avoid current flowthrough M1 and related motor areas, in a control experiment the twoelectrodes were placed over the occipito-temporo-parietal junction,centred between O1-P3 and O2-P4, respectively, according to the 10–20 system. Because tDCS functions in a bipolar way, in six cases theanode was placed over the right side and in seven cases over the leftside.

Functional magnetic resonance imaging (fMRI)

fMRI studies were conducted at 3 T (Magnetom TIM Trio, SiemensHealthcare, Erlangen, Germany) using a standard eight-channelphased array head coil. Subjects were placed supine inside themagnet bore and wore headphones for noise protection. Vitalfunctions were monitored throughout the experiment. Initially,anatomic images based on a T1-weighted 3D turbo fast low angleshot (FLASH) MRI sequence at 1 mm3 isotropic resolution wererecorded (repetition time (TR)=2250 ms, inversion time: 900 ms,echo time (TE)=3.26 ms, flip angle: 9°). For BOLD fMRI a multisliceT2*-sensitive gradient-echo echo-planar imaging (EPI) sequence(TR=2000 ms, TE=36 ms, flip angle 70°) at 2×2 mm2 resolutionwas used. Twenty two consecutive sections at 4 mm thickness in an

592 A. Antal et al. / NeuroImage 55 (2011) 590–596

axial-to-coronal orientation roughly parallel to the intercommisuralplane were acquired, covering the brain areas of interest.

The effects of the tDCS equipment on EPI raw image quality, i.e. thesignal-to-noise-ratio (SNR) and susceptibility artefacts, were assessedby recording functional raw images from one subject both withoutthe tDCS equipment and with the tDCS equipment (Fig. 3). In detail,we performed measurements under the following conditions: (i) notDCS equipment; (ii) electrodes attached to the subject's head;(iii) electrodes attached and connecting cable running through thecabin wall; (iv) electrodes attached and cable connected to thestimulator without stimulation; and (v) electrodes attached and cableconnected to the stimulator with stimulation. SNR was calculated forthree regions-of-interest (ROIs) by dividing the mean signal from theROI by the standard deviation of the background noise. ROIs werechosen to represent areas relatively far away from the tDCS electrodesand corresponding current flow as well as close to the tDCS electrodeplaced over M1. Accordingly, one ROI was located in left-hemisphericwhite matter above the ventricle, one in right-hemispheric whitematter anterior to the central sulcus, close to the M1 hand area, andone in right-hemispheric gray matter covering the omega-shaped

Fig. 3. (A) Axial cross-sections of the T1-weighted anatomical dataset of a single subjectequipment. (C) With electrodes attached to the subject's head. (D) With electrodes attachedcable connected to the stimulator without stimulation. (F)With electrodes attached and cablM1 hand area (arrowheads) and right-hemispheric orbita (arrows) are indicated. The positiothe tDCS electrodes (white circle), in white matter close to the tDCS electrode placed oveelectrode (red) are given. The corresponding SNR-values for a specific set-up are given nextare indicated by asterisks. Magnified views of the EPI raw images from (F) focusing on the

M1 hand area (Dechent and Frahm, 2003). Susceptibility artefacts,i.e. image distortions and signal dropouts, were evaluated by visualinspection.

Experimental protocol

The stimulation paradigm was implemented as a block design withtask and rest phases of 20 s each, repeated eight times. Five runs wereperformed with the following conditions task-phase: (1) anodal tDCS(anodal), (2) cathodal tDCS (cathodal), (3) finger tapping (ft), (4) fingertapping plus anodal tDCS (ft+anodal), and (5) finger tapping pluscathodal tDCS (ft+cathodal). The stimulation at rest (anodal andcathodal)was always performed at the beginning of the experiment andthe order of the stimulation was randomised. After this the ft, (ft+anodal) and (ft+cathodal) conditions were randomised and counter-balanced (5–4–4). Thirteen subjects participated in the main experi-ment, in all of the experimental conditions. Between the differentconditions there was a 2–5 minute break to avoid stimulation after-effects. According to the previous results at least a 3-minute stimulationduration is necessary to induce after-effects (Nitsche and Paulus, 2000).

and (B–F) corresponding T2*-weighted EPI raw images. (B) Without additional tDCSand connecting cable running through the cabin wall. (E) With electrodes attached ande connected to the stimulator with stimulation. The electrodes over the left-hemisphericns of the three ROIs chosen for SNR calculation in white matter relatively far away fromr M1 (yellow circle), and in gray matter covering the M1 hand area close to the tDCSto the respective row of EPI raw images. Susceptibility artifacts caused by the electrodesarea where the electrodes are located are given in the bottom row.

593A. Antal et al. / NeuroImage 55 (2011) 590–596

During finger tapping subjects were asked to move the index finger tothe thumb of the right hand. The requested motor performance wascued by anauditory signalwith a frequencyof 1 Hz. The executionof thetask was monitored visually by the experimenter. The auditory stimuliwere not presented in the condition inwhich no ftwas required (anodalonly and cathodal only conditions). In the ft+anodal and ft+cathodalconditions tDCS was applied only during the 20 s task-phase. tDCS wasapplied at 1.0 mA intensity, ramping up and down over the first and last2 s of the 20 s stimulation duration.

The control experiment inwhich theelectrodeswere placed over theoccipito-temporo-parietal junctionwas performed in a separate sessionat least 2 weeks apart from the main experiment. Three runs wereperformed with the following conditions task-phase: (1) finger tapping(ft), (2) finger tapping plus anodal tDCS (ft+anodal), and (3) fingertapping plus cathodal tDCS (ft+cathodal). The parameters of thestimulation and task were exactly the same as in the main experiment.

Data analysis

Functional data were preprocessed and analyzed with “BrainVoya-gerQX” (version 1.10.2., Brain Innovation,Maastricht, TheNetherlands).Preprocessing included 3Dmotion correction, slice scan time correction,linear trend removal and spatial smoothing with a Gaussian kernel4×4×4 mm³ full width at half maximum. Subsequently, functionaldatasets were co-registered to the anatomical dataset and transformedinto Talairach (TAL) space. Afterwards, using the BrainVoyager Volume-Of-Interest Analysis tool, a volume of interest (VOI) random effectsanalysis (RFX) was performed in the following way: First we identifiedVOIs belonging to the brain motor network by applying a RFX to the ftcondition at a corrected threshold of q(FDR)=0.05. In order toinvestigate whether the stimulation conditions without ft (anodal andcathodal conditions) had any effect, a one-way two-levels (anodal,cathodal) repeated measures RFX-ANOVA was performed on each VOI.Second, in order to identify whether the stimulation combined with fthad any effect with respect to the condition ft alone, a one-way three-levels (ft, ft+anodal, and ft+cathodal) repeatedmeasures RFX-ANOVAwas applied to each VOI. If significant differences were found after theRFX-ANOVA analysis, post hoc tests were used to show specificdifferences (paired t-tests).

In the control experiment with regard to the stimulation of theoccipito-temporo-parietal junction conditions, ft and ft+ stimulationwere regarded as the different predictors; therefore the functionalcontrast was calculated between these conditions. Ft+ stimulationcondition means that anodal and cathodal stimulations wereconsidered together, because i) this cortical area was chosen as acontrol to test possible non-specific effects of the stimulation; and ii)with regard to the ft a ‘target’ electrode position cannot be definedover the occipito-temporo-parietal junction, as we can do with M1stimulation. For statistical analysis a VOI RFX one-way two-levels (ft,ft+stimulation) repeated measures ANOVA was carried out.

Results

All of the subjects tolerated the stimulation well; none reportedside-effects during or after the stimulation. All of the subjects reportedlight itching under the electrodes that was not polarity dependent.The functional raw images were only very mildly affected byintroducing the tDCS equipment into the scanner environment(Fig. 3). Compared to the measurement without tDCS equipment,SNR was hardly reduced with decreases ranging from 3 to 8% for thedifferent ROIs and setups, even in the gray matter ROI in M1 targetedby tDCS. Interestingly, the ROIs close to the electrode were not moreaffected than the one far away from the electrode. Most importantly,SNR was identical regardless of whether the tDCS stimulator wasturned off or on, i.e. without or with direct current flow. Mildsusceptibility artefacts were observed under the frontal electrode,

however, not affecting the underlying brain tissue. The electrode overM1 did not cause any visible artefacts.

The identified motor network VOIs were the following (with theircorresponding x–y–z TAL cluster center and size): Left M1 (x=−34,y=−26, z=51; 569 voxels), SMA (x=−3, y=−7, z=51; 194voxels), left basal ganglia (x=−46, y=−24 z=5; 539 voxels), rightbasal ganglia (x=20, y=−3, z=10; 251 voxels) (Fig. 4).

Table 1 shows the results of the RFX-ANOVA analyses where first,it can be observed that neither anodal nor cathodal conditions hadany effect on the activation of the motor network VOIs. RespectiveBOLD time courses for left M1 are given in Fig. 4 (lower-left panel).Second, the RFX-ANOVA performed for the ft+ stimulation condi-tions demonstrated that the only region revealing a significant effectwas the SMA (F(2,24)=6.77; p=0.0046) (Fig. 5A and Table 1). Thepost hoc t-tests revealed that only the contrast ft+anodalN ft had aneffect (peak TAL x=−6, y=−13, z=53; t(12)=4.1, p=0.0015)(Figs. 5B and C). The contrast ft+cathodalN ftwas close to be signifi-cant (peak TAL x=−4, y=−13, z=51; t(12)=1.98, p=0.071).After the individual inspection of the data we observed that from 13subjects 12 showed a decreased BOLD response in SMA during the ft+anodal condition, however, during ft+cathodal condition only eightsubjects demonstrated reduced BOLD activity (Fig. 6). While the effectwas clear on the single subject level in these eight participants, thediminution of the functional activity was not significant on the grouplevel.

There was no differential BOLD signal change in the brainparenchyma under the stimulation electrode (M1) using eitherstimulation conditions. Similarly, the electrical stimulation of theoccipito-temporo-parietal area in the control experiment resulted inno significant changes between ft and ft+stimulation conditions(Table 2).

Discussion

The data presented here show a decrease in BOLD activity in SMAwhen the M1, during a simple finger tappingmovement, is stimulatedby anodal, “excitatory” current flow. Cathodal, ‘inhibitory’ stimulationof the M1 did not result in significant change in BOLD response,however, a tendency toward decreased activity was observed on thegroup level. Further inspection of single subject data revealed asignificant effect in eight out of the 13 subjects. In contrast, tapping-induced BOLD activity in the M1 itself is unaffected by tDCSinterventions. Normally, excitatory and inhibitory aftereffects withanodal or cathodal tDCS are derived from MEP measurements withtDCS applied at rest (Nitsche and Paulus, 2000). In human studies asingle TMS pulse over the M1 with a standardized intensity, induces aMEP with a target size of about 1 mV, the amplitude of which iscompared between pre- and post-tDCS intervention. Post-tDCSincrease of MEP sizes with anodal and decrease with cathodalstimulation (Nitsche and Paulus, 2000) could be modified by motorand cognitive task performance. For example, with ball compressionmovements during tDCS the excitatory anodal stimulation does notincrease MEP amplitudes but decreases them (Antal et al., 2007).

Since there are no sufficient direct projections from the SMA to thespinal cord, the TMS routine paradigms over M1 as described beforecannot be used for direct comparison between BOLD and neurophys-iology. Only indirect measures are achievable by targeting either SMAwith repetitive TMS (e.g. Hamada et al., 2009) or premotor cortexwith tDCS (e.g. Boros et al., 2008) and measuring its influence on M1output. Instead fMRI as used here offers the unique possibility toexplore remote effects in connected areas. In the present paper wehave seen decreased BOLD response in SMA during tDCS. The self-initiated movement used in the present paradigm is expected to begenerated preferentially in SMA (Nachev et al., 2008). We did not seeany change in the BOLD activity in M1 during stimulation, however,we know from early animal work that DC stimulation may increase or

Fig. 4. Activation maps showing the RFX during ft over the averaged T1-weighted dataset. The identified VOIs corresponding to the motor network and the auditory cortex areindicated: supplementary motor area (SMA), primary motor cortex (M1), left basal ganglia (lBG), right basal ganglia (rBG), and left auditory cortex (lAC). In the lower-left panel thex-axis shows the time (in seconds) of the averaged blocks starting four seconds before the initiation of the task phase (left white vertical bar) ending up twelve seconds after theinitiation of the rest phase. The error bars show the standard error of the mean.

594 A. Antal et al. / NeuroImage 55 (2011) 590–596

decrease or even silence firing neurons in M1 (Bindman et al., 1964).Thus, neural activity in the stimulated area can be influenced by tDCSand physiological connections may amplify selectively the activitychanges in projection areas. Besides this, we have to consider thatdirect membrane de- or hyperpolarization may be less energyconsuming than alteration of the activity of remote projection areasvia amplification of firing rates of diversely connected fibres, and thatBOLD imaging might not be equally sensitive to monitor these twoprocesses. It is also suggested that the BOLD response may reflectrather synaptic than spiking activity (Viswanathan and Freeman,2007). Consequently, our results may reflect the change in synapticactivity after anodal tDCS, an issue requiring further investigation.

With regard to the layer-specific targeting of tDCS, recentexperimental data in rat slices indicate that cortical neuron morphol-ogy, type and orientation relative to applied electrical fields are keyvariables determining sensitivity to sub- and suprathreshold electricalstimulation in the brain (Radman et al., 2009). It is has been suggestedthat the soma of the pyramidal neurons in layer V is themost sensitiveto polarization by optimally oriented subthreshold fields, but

Table 1The results of the statistical analysis with regard to the M1 stimulation.

One-way repeated measures ANOVA

VOIs Stimulation only ft+stimulation

F-value(1,12) p-value F-value(2,24) p-value

SMA 1.6 0.21 6.77 0.005Left M1 2.1 0.16 1.14 0.33Left basal ganglia 1.7 0.22 1.4 0.26Right basal ganglia 1.13 0.35 2.1 0.15

probably the neuronal activity of other layers is affected, too.Furthermore, it is not clear whether these animal data can directlybe translated to the human M1.

Although we did not see any differential response in the premotorcortex, it is possible that in addition to M1, the left caudal dorsalpremotor cortex (cdPM)was also directly stimulated and resulted in aremote change in activity in SMA. Because we have used relativelylarge electrodes, this cortical area was probably covered by thestimulating electrode. Since the cdPM cortex is located superficially atthe crown of the precentral gyrus and due to the radial orientation ofthe pyramidal cells, with respect to the hemispheric surface, it createsan optimal surface for polarization. Therefore it cannot be excludedthat the cdPM was targeted by tDCS as well. Further studies usingfocal electrodes could better clarify this question.

In a recent study a visually cued serial reaction time task wasperformed in the scanner before and after 10 min of 1 mA tDCSapplied to the left M1 (Stagg et al., 2009). Anodal tDCS led to short-lived activation increases in the left M1 and in the SMA. Cathodal tDCSresulted in an increase in activation in the contralateral M1 and in thedorsal PM cortex and increased functional connectivity between theseareas and the stimulated left M1. However, in a previous paper(Baudewig et al., 2001) in which 5 min tDCS was applied before afinger-tapping task, cathodal tDCS resulted in a 57% decrease ofactivated pixels in the SMA, but induced no change in the M1. AnodaltDCS yielded a nonsignificant 5% increase of activated pixels with noregional differences. Similarly in our study anodal and cathodal tDCSover M1 applied during a motor task did not lead to any detectablesignal differences in M1 when compared to the motor task withoutstimulation. Furthermore, we did not see any detectable signal changewhen the stimulationwas applied without amotor task. However, ourstudy is different from previous studies with regard to the stimulation

Fig. 5. (A) Activationmaps showing the one-way repeatedmeasures ANOVAwith ft andft+anodal and ft+cathodal as levels. (B) The post hoc paired t-tests revealed that onlyft+anodal compared with ft has an effect (peak TAL x=−6,y=−13,z=53; t(12)=4.1, p=0.0015). Both panels (A) and (B) show the traversal slice at TAL z=53. (C) Thex-axis shows the time (in seconds) of the averaged blocks in the SMA starting fourseconds before the initiation of the task phase (left white vertical bar) ending up twelveseconds after the initiation of the rest phase. The error bars show the standard error ofthe mean.

Fig. 6. Individual beta plot values of the SMA for ft+anodal− ft (top) and ft+cathodal− ft(bottom).

595A. Antal et al. / NeuroImage 55 (2011) 590–596

duration: we have used very short stimulation duration and did notintend to induce any after-effect. In contrast to this, the aim of theprevious studies was to induce short or longer lasting after-effects,because they intended to test the effectiveness of the stimulationusing a given motor task.

Nevertheless, it is well documented by PET that the application of1 mA tDCS for 10 min results in measurable alteration in regionalcerebral blood flow (rCBF) at rest after tDCS (Lang et al., 2005). Theabsence of corresponding BOLD signal changes therefore indicatesthat PET may be a more sensitive technique, and better suited thanfMRI to address weak differences in activation between the tDCS andrest conditions (Ito et al., 2005). By measuring rCBF at rest and duringfinger movements after tDCS it was found that anodal and cathodaltDCS induced widespread increases and decreases in rCBF in corticaland subcortical areas (in the left M1, right frontal pole, and rightprimary sensorimotor cortex), when compared to sham tDCS at rest(Lang et al., 2005). These changes in rCBF remained stable throughoutthe 50-minute period of PET scanning. With regard to the concurrentapplication of tDCS and fMRI, it has been demonstrated that at restduring anodal stimulation over the M1, a direct facilitatory effect onthe underlying cortex and remote BOLD signal increases in the leftSMA and the right posterior parietal cortex can be observed, but onlyin the fourth out of 4 consecutive stimulation sessions (Kwon et al.,2008). Thus, more sophisticated techniques may improve sensitivityof fMRI further when compared with PET. On the other hand,increasing the intensity or duration of the stimulation might alsoresult in different effects. Indeed, it was lately observed that the BOLDsignal evoked by grasping hand movements in a consecutiveparadigm was increased in the sensorimotor area after 20 min anodaltDCS when compared to baseline (Jang et al., 2009). The authorsconcluded that a longer application of tDCS and a simpler motor taskmight be better to detect changes in cortical activation than the morecomplex tasks used by Baudewig et al. (2001). In the present studywe

Table 2The results of the statistical analysis with regard to the electrode position over theoccipito-temporo-parietal area.

One-way repeated measures ANOVA

VOIs Control experiment

F-value(1,12) p-value

SMA 1.5 0.21Left M1 2.1 0.16Left basal ganglia 2.5 0.13Right basal ganglia 1.6 0.22

596 A. Antal et al. / NeuroImage 55 (2011) 590–596

have also preferred to use a paradigm that does not produce anydifference at behavioral level in order to see only the possible DC-induced changes on the functional level. However, one of thelimitations of the present study might be that the applied paradigmwas too simple and automatic and the motor system could effortlesslyrespond to the demand introduced here, thereforewemight not be ableto see small changes elicited by the stimulation. Second, we have tomention that theVOImethodweused in this studymightbenot thebestto determine the efficacy of the tDCS on the BOLD activity. Theanatomical shifts that canbeproducedby the stimulationwithin theVOIcannot be detected with this method and also outside the VOIs theeffects cannot be observed.

Until recently, a combination of TMS and fMRI dominated thefield ofconcurrent applications of neuroimaging and non-invasive stimulationmethods (Bohning et al., 1999; Baudewig et al., 2001; Bestmann et al.,2006). Indeed, recent concurrent TMS–fMRI studies can illustrate howthis combined technique provides exceptional insights into causalinteractions amongbrain regions in human subjects. fMRI can detect thespatial topography of local and remote TMS effects applied overdifferent cortical areas (for a review see: Bestmann et al., 2008). Evensingle pulse TMS toM1 or PMd can evoke significant activity changes inremote regions of the motor system. Recently, even the inhibitory andfacilitatory effects of paired-pulseTMShavebeenmonitoredusingBOLDfMRI (Baudewig et al., 2009). Furthermore, the novel combination ofTMS with arterial spin labeling offers the possibility to investigate theimmediate and enduring after-effects of repetitive TMS stimulation onrCBF (Moisa et al., 2010). The combination of tDCS and fMRI has beentechnically challenging because of electrode artifacts and the remoterisk of suddenelectrodeheating. As afirst result and in agreementwith aprevious study by Kwon et al. (2008), the combination of tDCS andfunctionalMRI of thehumanbrain canbeperformed as a safeprocedure,even when used here for the first time in a 3 T environment. Potentialtechnical challenges, such as possible susceptibility-related echo-planarimaging artifacts under the electrode and a lower SNR, were taken intoconsideration. However, careful technical design and application of theequipment lead to a minimum amount of negative effects and ensuredsuccessful investigations.

Conclusion

In this studyweusedbothcathodal and anodal tDCSduring fMRI.Weconclude that the observation of the BOLD signal attenuation to a largedegree fits into a concept of tDCS acting as a modulator. In our study,while the effect of cathodal stimulation was somewhat variable, anodaltDCS resulted in an attenuation of the BOLD response in the SMAwell inline with neurophysiological behavior during activation. Indeed, whentDCSwas applied during a simplemotor task requiring fist-movements,the amplitudes of the MEPs were dramatically reduced, independentlyfrom the polarity of stimulation (Antal et al., 2007). This behavior underactivation also applies for repetitive TMS (Todd et al., 2009).

Until now, combined tDCS and fMRI experiments were imple-mented in a sequential manner to investigate any enduring BOLDsignal changes in the brain (e.g. Baudewig et al., 2001). Although theBOLD response is an indirect measure of neuronal activity, thecombination of fMRI and concurrent tDCS may allow a more directvisualization of the electrical stimulation-induced changes in brainactivity with high spatial resolution and the possibility to chart howtDCS modifies ongoing brain activations.

Acknowledgments

This study was funded by the Research program of the UniversityGoettingen (Forschungsförderprogramm 2009 — Startförderung;AA), the Rose Foundation (RP, WP), the Bernstein Center forComputational Neuroscience (01GQ0782) (WP), and the VolkswagenFoundation (PD).

References

Antal, A., Paulus, W., 2008. Transcranial direct current stimulation and visualperception. Perception 37, 367–374.

Antal, A., Terney, D., Poreisz, C., Paulus, W., 2007. Towards unravelling task-relatedmodulations of neuroplastic changes induced in the human motor cortex. Eur. J.Neurosci. 26, 2687–2691.

Baudewig, J., Nitsche, M.A., Paulus, W., Frahm, J., 2001. Regional modulation of BOLDMRI responses to human sensorimotor activation by transcranial direct currentstimulation. Magn. Reson. Med. 45, 196–201.

Baudewig, J., Schmidt-Samoa, C., Dechent, P., 2009. BOLD imaging of inhibition andfacilitation induced by paired-pulse transcranial magnetic stimulation: feasibilityand reproducibility. Neuroimage 47, S150.

Bestmann, S., Oliviero, A., Voss, M., Dechent, P., Lopez-Dolado, E., Driver, J., Baudewig, J.,2006. Cortical correlates of TMS-induced phantom hand movements revealed withconcurrent TMS–fMRI. Neuropsychologia 44, 2959–2971.

Bestmann, S., Ruff, C.C., Blankenburg, F., Weiskopf, N., Driver, J., Rothwell, J.C., 2008.Mapping causal interregional influences with concurrent TMS–fMRI. Exp. Brain Res.191, 383–402.

Bindman, L.J., Lippold, O.C., Redfearn, J.W., 1964. The action of brief polarizing currentson the cerebral cortex of the rat (1) during current flow and (2) in the production oflong-lasting after-effects. J. Physiol. 172, 369–382.

Bohning, D.E., Shastri, A., McConnell, K.A., Nahas, Z., Lorberbaum, J.P., Roberts, D.R.,Teneback, C., Vincent, D.J., George, M.S., 1999. A combined TMS/fMRI study ofintensity-dependent TMS over motor cortex. Biol. Phychiat. 45, 385–394.

Boros, K., Poreisz, C., Münchau, A., Paulus, W., Nitsche, M.A., 2008. Premotortranscranial direct current stimulation (tDCS) affects primary motor excitabilityin humans. Eur. J. Neurosci. 27, 1292–1300.

Dechent, P., Frahm, J., 2003. Functional somatotopy of finger representations in humanprimary motor cortex. Hum. Brain Mapp. 18, 272–283.

Fregni, F., Boggio, P.S., Lima, M.C., Ferreira, M.J., Wagner, T., Rigonatti, S.P., Castro, A.W.,Souza, D.R., Riberto, M., Freedman, S.D., Nitsche, M.A., Pascual-Leone, A., 2006. Asham-controlled, phase II trial of transcranial direct current stimulation for thetreatment of central pain in traumatic spinal cord injury. Pain 122, 197–209.

Hamada, M., Hanajima, R., Terao, Y., Okabe, S., Nakatani-Enomoto, S., Furubayashi, T.,Matsumoto, H., Shirota, Y., Ohminami, S., Ugawa, Y., 2009. Primary motor corticalmetaplasticity induced by priming over the supplementary motor area. J. Physiol.587, 4845–4862.

Hellwag, C.F., 1802. Erfahrungen über die Heilkräfte des Galva-nismus, und Betrach-tungen über desselben chemische und physiologische Wirkungen. Mit 1 Kupfer.Und Beobachtungen bey der medicinischen Anwendung der Voltaischen Säule vonMaximilian Jacobi. Hamburg: Per-thes.

Ito, H., Ibaraki,M., Kanno, I., Fukuda, H.,Miura, S., 2005. Changes in cerebral blood flow andcerebral oxygen metabolism during neural activation measured by positron emissiontomography: comparisonwith blood oxygenation level-dependent contrastmeasuredby functional magnetic resonance imaging. J. Cereb. Blood Flow Metab. 25, 371–377.

Jang, S.H., Ahn, S.H., Byun, W.M., Kim, C.S., Lee, M.Y., Kwon, Y.H., 2009. The effect oftranscranial direct current stimulation on the cortical activation by motor task inthe human brain: an fMRI study. Neurosci. Lett. 460, 117–120.

Kwon, Y.H., Ko, M.H., Ahn, S.H., Kim, Y.H., Song, J.C., Lee, C.H., Chang, M.C., Jang, S.H.,2008. Primary motor cortex activation by transcranial direct current stimulation inthe human brain. Neurosci. Lett. 435, 56–59.

Lang,N., Siebner,H.R.,Ward,N.S., Lee, L., Nitsche,M.A., Paulus,W., Rothwell, J.C., Lemon,R.N.,Frackowiak, R.S., 2005. How does transcranial DC stimulation of the primary motorcortexalter regional neuronal activity in thehumanbrain?Eur. J.Neurosci. 22, 495–504.

Liebetanz, D., Fregni, F., Monte-Silva, K.K., Oliveira, M.B., Amâncio-dos-Santos, A.,Nitsche, M.A., Guedes, R.C., 2006. After-effects of transcranial direct currentstimulation (tDCS) on cortical spreading depression. Neurosci. Lett. 398, 85–90.

Moisa, M., Pohmann, R., Uludağ, K., Thielscher, A., 2010. Interleaved TMS/CASL:comparison of different rTMS protocols. Neuroimage 49, 612–620.

Nachev, P., Kennard, C., Husain, M., 2008. Functional role of the supplementary and pre-supplementary motor areas. Nat. Rev. Neurosci. 9, 856–869.

Nitsche, M.A., Paulus,W., 2000. Excitability changes induced in the humanmotor cortexby weak transcranial direct current stimulation. J. Physiol. 527, 633–639.

Nitsche, M.A., Paulus, W., 2001. Sustained excitability elevations induced bytranscranial DC motor cortex stimulation in humans. Neurology 57, 1899–1901.

Nitsche, M.A., Cohen, L.G., Wassermann, E.M., Priori, A., Lang, N., Antal, A., Paulus, W.,Hummel, F., Boggio, P.S., Fregni, F., Pascual-Leone, A., 2008. Transcranial directcurrent stimulation: state of the art 2008. Brain Stimul. 1, 206–223.

Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburghinventory. Neuropsychologia 9, 97–113.

Priori, A., Berardelli, A., Rona, S., Accornero, N., Manfredi, M., 1998. Polarization of thehuman motor cortex through the scalp. NeuroReport 9, 2257–2260.

Radman, T., Ramos, R.L., Brumberg, J.C., Bikson, M., 2009. Role of cortical cell type andmorphology in subthreshold and suprathreshold uniform electric field stimulationin vitro. Brain Stimul. 2, 215–228.

Stagg, C.J., O'Shea, J., Kincses, Z.T., Woolrich, M., Matthews, P.M., Johansen-Berg, H.,2009. Modulation of movement-associated cortical activation by transcranial directcurrent stimulation. Eur. J. Neurosci. 30, 1412–1423.

Todd, G., Rogasch, N.C., Flavel, S.C., Ridding, M.C., 2009. Voluntary movement andrepetitive transcranial magnetic stimulation over human motor cortex. J. Appl.Physiol. 106, 1593–1603.

Viswanathan, A., Freeman, R.D., 2007. Neurometabolic coupling in cerebral cortexreflects synaptic more than spiking activity. Nat. Neurosci. 10, 1308–1312.

Webster, B.R., Celnik, P.A., Cohen, L.G., 2006. Noninvasive brain stimulation in strokerehabilitation. NeuroRx 3, 474–481.