Abstract

We examined the stimulus–response profile during single-pulse transcranial magnetic stimulation (TMS) by measuring motor-evoked potentials (MEPs) with electromyographic monitoring and hemodynamic responses with functional magnetic resonance imaging (fMRI) at 3 Tesla. In 16 healthy subjects, single TMS pulses were irregularly delivered to the left primary motor cortex at a mean frequency of 0.15 Hz with a wide range of stimulus intensities. The measurement of MEP proved a typical relationship between stimulus intensity and MEP amplitude in the concurrent TMS-fMRI environment. In the population-level analysis of the suprathreshold stimulation conditions, significant increases in hemodynamic responses were detected in the motor/somatosensory network, reflecting both direct and remote effects of TMS, and also the auditory/cognitive areas, perhaps related to detection of clicks. The stimulus–response profile showed both linear and nonlinear components in the direct and remote motor/somatosensory network. A detailed analysis suggested that the nonlinear components of the motor/somatosensory network activity might be induced by nonlinear recruitment of neurons in addition to sensory afferents resulting from movement. These findings expand our basic knowledge of the quantitative relationship between TMS-induced neural activations and hemodynamic signals measured by neuroimaging techniques.

Introduction

Transcranial magnetic stimulation (TMS) is a noninvasive technique that stimulates a localized brain region underneath the coil and thereby modulates brain activity. Single-pulse TMS can temporarily affect motor, sensory or cognitive behavior, and repetitive TMS (rTMS) protocols such as theta burst stimulation (Huang et al. 2005) can induce long-lasting plastic changes. Because of these unique properties, TMS has been widely applied to basic neuroscience research and various clinical settings. Nevertheless, the actions of TMS, presumptively involving both excitatory and inhibitory neural circuits, are not fully understood. A recent study investigated changes in neuronal activity and local field potentials simultaneously with tissue oxygenation in anesthetized cats (Allen et al. 2007). This study clearly demonstrated that TMS-induced local neuronal changes, either increases or decreases of activity, resulted in corresponding hemodynamic changes, and promised the helpfulness of hemodynamic measurements for monitoring the actions of TMS.

In fact, the combined use of TMS with positron emission tomography (PET) (Fox et al. 1997; Paus et al. 1997; Siebner et al. 1998, 2001; Speer et al. 2003), single photon emission computed tomography (Okabe et al. 2003), or functional magnetic resonance imaging (fMRI) (Bohning et al. 1998, 1999; Baudewig et al. 2001; Bestmann et al. 2004) has been emerging in the field of noninvasive human brain mapping. The stimulus–response relationship between various stimulus parameters and evoked hemodynamic signals is important for the interpretation of findings from simultaneous TMS-imaging studies. Several studies have already addressed intensity-dependent effects on brain activity, by incorporating more than 2 different levels of stimulus intensity into the experimental paradigm (Baudewig et al. 2001; Nahas et al. 2001; Bestmann et al. 2003; Speer et al. 2003; Fox et al. 2006). Peculiarly, at first glance, the directly stimulated area in these studies consistently requires a stronger stimulus for the induction of significantly increased brain activation measurable with neuroimaging techniques than do remote, nonstimulated areas. In response to TMS to the primary motor cortex (M1), for example, activity in the supplementary motor areas (SMA) is increased during both subthreshold and suprathreshold TMS, whereas activity in the directly stimulated M1 is induced only by suprathreshold TMS. Although these observations are extremely interesting, they might reflect complex interactions between intensity, frequency and train length rather than pure effects of intensity, as almost all intensity-response studies have employed rTMS paradigms. Evidence suggests that the effects of rTMS on brain activity may last for days (Hayashi et al. 2004), and thus caution must be paid to interpret brain activity changes from repeated runs of rTMS protocols. Meanwhile, Komssi and colleagues recorded electroencephalographic (EEG) responses immediately after delivery of single TMS pulses to the M1 at various stimulus intensities (Komssi et al. 2004). By taking advantage of the fine temporal resolution of the technique, the authors were able to demonstrate both linear and nonlinear aspects in the intensity-response profile. However, the neural correlates of the source(s) of these brain responses remain to be specified.

The purpose of the present study was to investigate the stimulus–response profile during single-pulse TMS in the directly stimulated and remote regions in the whole brain. Hemodynamic signals and motor-evoked potentials (MEPs) in response to single TMS pulses were sampled at a wide range of stimulus intensities. To achieve online monitoring of surface electromyography (EMG) during concurrent TMS-fMRI, we employed the stepping stone sampling (SSS) method developed for artifact-diminished acquisition of EEG during hemodynamic imaging (Anami et al. 2003). The hypothesis was that single-pulse TMS-induced changes in hemodynamic signals would show both linear and nonlinear components as suggested by a previous TMS-EEG study (Komssi et al. 2004). The present TMS-fMRI study was expected to identify the sources of these TMS-induced linear and nonlinear stimulus–response profiles in a spatially segregated manner. A preliminary result of this study has been published only in abstract form.

Materials and Methods

Subjects

Nineteen healthy adults (mean age = 31.4 years; range 24–39 years) were initially recruited. None of the subjects reported any history of neuropsychiatric disorders, especially epilepsy. All subjects were right handed as judged by Edinburgh handedness inventory. The study protocol was approved by the Kyoto University Graduate School and Faculty of Medicine, Ethics Committee. The subjects were fully informed about the experimental procedure, and all of them participated in the experiment after giving written informed consent.

TMS and EMG Monitoring

The optimal site at which TMS evoked a maximal motor response in the right abductor pollicis brevis (APB) muscle (“motor hot-spot”) was identified for each subject while sitting comfortably on a chair outside the scanner. The scalp position corresponding to the motor hot spot was marked with a felt-tip pen. We used the APB in this experiment because the cortical representation of the thumb is larger than that of the other fingers (Penfield and Boldrey 1937). We thought that this fact should help us to stimulate the cortical representation of the same finger consistently and thereby allow us to measure reliable fMRI signals during the whole fMRI session.

The subject then lay supine on the bed of a 3-Tesla whole-body MRI scanner equipped with a circular polarization head coil (Siemens Magnetom Trio, Erlangen, Germany). An MRI-compatible figure-of-8 TMS coil with an outer-wing diameter of 70 mm (MR coil, Magstim, Witland, Wales, UK) was positioned tangentially to the scalp at the marked site. The orientation of the TMS coil was approximately 45° away from the medial–lateral axis. The TMS coil was connected to a stimulator (SuperRapid, Magstim, Witland, UK) via a 10-m cable running through a wave guide tube appropriate for radiofrequency wave filtering. The stimulator produced biphasic electrical pulses of approximately 250-μs duration and a rise time of 50 μs.

For EMG recording, we employed a method originally developed for EEG recording during fMRI scanning (Anami et al. 2003). MEPs were recorded from the right APB. Silver/silver chloride surface electrodes with shielded plates and cables were placed over the right thenar eminence with an interelectrode distance of 3 cm. A ground electrode was placed on the dorsal surface of the right wrist. EMG signals were fed to a digital amplifier (SynAmps; Neuroscan, Sterling, VA) through a radiofrequency filter. EMG data were sampled at a digitization rate of 1 kHz with amplitude resolution of 0.336 μV/bit and a dynamic range of 22 mV. As the SSS method requires precise temporal consistency between the timing of EMG sampling and that of gradient pulses for MRI acquisition, the SynAmps amplifier was externally driven by the clock of the MRI scanner. For this purpose, the clock frequency was downsampled from the original 10 MHz to 10 kHz by a custom-made clock divider (CD5; Physio-Tech, Tokyo, Japan). Furthermore, a trigger pulse from the scanner was sent to the clock divider to synchronize the onset of EMG measurement and MRI acquisition. The EMG data were monitored and recorded with a low-pass filter of 2 kHz throughout the experiment (Fig. 1A).

Figure 1.

(A) An example EMG recording during concurrent TMS-fMRI using the “SSS” method. Imaging artifacts are overlain onto the EMG record during fMRI scanning, as subtraction of imaging artifacts was not performed. Even without artifact subtraction, MEPs are much larger than scanning artifacts. (B) An example of the functional imaging volume acquired with the SSS imaging sequence just after a single TMS pulse.

Figure 1.

(A) An example EMG recording during concurrent TMS-fMRI using the “SSS” method. Imaging artifacts are overlain onto the EMG record during fMRI scanning, as subtraction of imaging artifacts was not performed. Even without artifact subtraction, MEPs are much larger than scanning artifacts. (B) An example of the functional imaging volume acquired with the SSS imaging sequence just after a single TMS pulse.

In the MRI environment, the position of the TMS coil was adjusted while stimulation was delivered every 5 s to elicit abduction of the right thumb. After stable production of MEPs was ascertained with EMG monitoring, the TMS coil was fixed immobile to the scanner bed with a custom-made coil holder made of polyetheretherketone plastic. To minimize head motion during the scanning, foam pads or vacuum cushions were utilized. After the subject's head was positioned at the gantry center, the resting motor threshold (rMT) was defined individually as the percentage of stimulator output that elicited MEPs of >50 μV peak-to-peak amplitude in the APB at rest in more than 5 of 10 successive trials (Rossini et al. 1994).

Experimental Paradigm

During scanning, subjects were instructed to relax and not to fall asleep. Vision was not constrained. In each fMRI run, 20 TMS pulses were delivered at a mean frequency of ∼0.15 Hz. Stimulus onset asynchrony was semirandomized (5.4 or 8.1 s). The stimulation timing was controlled by Presentation software (Neurobehavioral systems, Albany, CA) on a personal computer synchronized with the MRI scanner via transistor–transistor logic pulses converted from the default optic signals of the scanner. To avoid image degradation, TMS pulses were delivered in the middle of the 200-ms scan delay periods, during which no gradient magnetic fields or radiofrequency pulses were generated for MRI acquisition. Twelve out of 19 subjects received TMS stimulation from 30% to 95% of the machine output at 5% steps. The stimulus intensity remained the same in each run; one to 2 fMRI runs were prescribed per TMS intensity. With this protocol, however, suprathreshold intensity was not sufficiently covered in some subjects because of technical reasons discussed later. In the remaining 7 subjects, therefore, stimulation intensity up to 110% of the original machine output was used. By using a special controller provided by the manufacturer, the maximum machine output of the TMS stimulator was temporarily reset to 110% of the default machine output (enhanced mode). For consistency, the physical intensity of TMS stimulation is always defined as the percentage of the default machine output (in the range 30–110%) in this article. In the latter group of subjects, stimulus intensity was varied from 30% of the machine output to <80% of the rMT in 10% steps and from 80% of the rMT to 110% of the machine output in 5% steps. The stimulus schedule was either an increasing–decreasing sequence or a decreasing–increasing sequence, and was counterbalanced across subjects.

Image Acquisition

The SSS scheme was utilized to measure hemodynamic signals with a blip-type echo planar imaging (EPI) sequence with following parameters: repetition time (TR) = 2.7 s, intervolume acquisition delay = 200 ms, echo time (TE) = 30 ms, flip angle (FA) = 90°, 64 × 64 matrix, 30 slices, field of view (FOV) = 192 mm, 3 × 3 × 3.75-mm voxel size, bandwidth = 1086 Hz). In brief, the timing of gradient pulses was carefully designed so that electrophysiological signals could be sampled every 1 ms (for a digitization rate at 1 kHz), in the designated periods, whereas the induced electric currents due to magnetic-field switching stayed around the baseline level (Anami et al. 2003). The scanner and the EMG amplifier were perfectly synchronized as described above. In addition, the intervolume acquisition delay allowed us to acquire EPI data without interference from induced electromagnetic fields or vibration of the TMS coil following pulse delivery (Fig. 1B). Each subject underwent 16–21 fMRI runs, and each run consisted of 70 functional volumes (total scanning time = 3 min, 9 s).

For anatomical registration, T2-weighted turbo spin echo images were obtained in the same space with the functional images (TR = 7080 ms, TE = 73 ms, FA = 122°, FOV = 192 mm, matrix = 256 × 256, voxel size = 0.75 × 0.75 × 3.75 mm). T1-weighted 3-dimensional structural images were also acquired with a magnetization-prepared, rapid-gradient echo sequence (TR = 2,000 ms, TE = 4.38 ms, FA = 8°, FOV = 240 mm, matrix = 256 × 256, voxel size = 1 × 1 × 1 mm).

EMG and Image Data Analysis

To assess the cortico-motoneuronal output, the peak-to-peak amplitude of MEP for each stimulation was measured and the data were averaged for each session (Scan 4, Neuroscan, Sterling, VA). The imaging data were preprocessed and analyzed with SPM5 (Wellcome Department of Imaging Neuroscience, UCL, London, UK) implemented on Matlab7 (MathWorks, Inc., Natick, MA). The first 4 volumes in each experimental run were discarded to allow for T1 equilibrium effects. The remaining functional images were corrected for differences in slice acquisition timing and spatially realigned to the first image of the first run to adjust for head motion. The realigned images were spatially normalized to fit into a Montreal Neurological Institute template based on the standard stereotaxic coordinate system. Subsequently, all images were smoothed with an isotropic Gaussian kernel of 6-mm full-width at half-maximum.

The events representing onsets of TMS were modeled as a main regressor for the first-level multiregression analysis. This analysis was performed for each subject to test the correlation between hemodynamic signal changes and a train of delta functions (representing the onsets of TMS) convolved with the canonical hemodynamic response function and its temporal derivative. Six parameters representing the head motion calculated in the realignment step were included in the design matrix as covariates of no interest. Global signal normalization was performed between runs. Low frequency noise was removed using a high-pass filter with a cut-off of 128 s, and serial correlations were adjusted using an autoregression model. By applying linear contrasts to the parameter estimates, mean effect images reflecting the magnitude of correlations between the hemodynamic signals and the behavioral model of interest were computed. The summary images were used for the subsequent second-level, random-effect model analysis.

Group-level statistical parametric maps (SPM) were produced by performing one-sample t-tests for the contrast of the strongest, suprathreshold stimulation (95–110% of machine output depending on the availability) and the weakest stimulation (30% of machine output in all subjects). In the group-level SPM, the threshold was initially set at a voxel-wise height-level P < 0.001, and activity with a cluster size of P < 0.05 corrected for multiple comparisons was regarded significant. The cytoarchitectonic nomenclature of significant brain activity was identified according to the SPM anatomy toolbox (Eickhoff et al. 2005) when applicable.

To assess intensity-dependent modulation of hemodynamic signals, brain activity was calculated from representative brain regions of interest determined on an a priori basis. The percent signal changes in hemodynamic signals was estimated in each region in each subject by setting up 5-mm radius spherical VOIs with the Marseille region of interest toolbox (http://marsbar.sourceforge.net/). The VOI-based method was chosen to assess both linear and nonlinear aspects of intensity-dependent modulation of brain responses. The VOIs were set up according to the individual's SPM from the contrast between the maximum and the minimum intensity conditions. Hence, subjects who did not reveal M1 activity in this contrast, known as “nonresponders” (Fox et al. 2006), were excluded from the group analysis. The VOI-based hemodynamic signal changes were then sampled from 9 representative areas, including the directly stimulated area (left M1), remote motor areas (SMA, cingulate motor areas, and bilateral dorsal lateral premotor cortex), somatosensory areas (left primary somatosensory cortex and bilateral second somatosensory cortices), bilateral primary auditory cortices, left thalamus, and the right cerebellum. Considering the size of the VOIs and the spatial resolution of the experiment, the midline structures (SMA and cingulate zones) should include data from both hemispheres. The left dorsal lateral premotor cortex (PMd) in the precentral gyrus and primary somatosensory cortex (S1) in the postcentral gyrus are very close to the left M1 and likely to be under the influence of direct TMS stimulation to some extent. Additionally, the left PMd and S1 have direct anatomical connections with the left M1, and the left S1 should also receive sensory afferents in the suprathreshold conditions. Therefore, interpretation of the VOI-based data sampled from the left PMd and S1 requires caution as the data may reflect complex interactions among these factors.

The signal changes were first evaluated as a function of the physical intensity (% of default machine output) of TMS. A repeated-measures analysis of variance (rmANOVA) was performed with a nonsphericity correction (Greenhouse–Geisser) having the stimulus intensity (bin width = 10% of machine output) as a within-subject variable. Pair-wise comparisons were performed to test if activity for each bin of the stimulus intensity was significantly greater than the activity for the minimum stimulation (30% machine-output bin). In this analysis, the 100% machine-output bin was assessed with a separate rmANOVA, because data were available from only 7 subjects.

Furthermore, TMS intensity was recalculated for each run for each individual by taking individual's rMT values into account. Percent changes in hemodynamic signals were assessed with regard to this rMT-adjusted, physiological intensity of TMS. These stimulus–response profiles were analyzed at 2 levels: one included both sub- and suprathreshold intensities (global-level analysis), whereas the other only included subthreshold intensity (subthreshold-level analysis). At the global level, curve fitting was performed with a sigmoid function (Boltzmann) to capture the nonlinear stimulus–response relationship between all levels of physiological TMS intensity and the hemodynamic signals (Origin7.5, OriginLab, Northampton, MA). For this purpose, the data were first pooled across all subjects. Next, data from each individual were fitted with both sigmoid and linear functions. To avoid irregular fitting, the point of inflection of the sigmoid fitting was constrained to fall within ±10 of the value obtained in the pooled data analysis. Each of these fitting analyses yielded the coefficient of determination as a rough measure of goodness of fit. To find which fitting could better explain the stimulus–response relationship in each area, the coefficient of determination was statistically compared across the 2 fitting analyses (Wilcoxon signed-rank test). At the subthreshold level, hemodynamic responses to TMS with subthreshold intensity (operationally defined as <80% rMT) were separately analyzed in each area. At this level, only linear fitting was applied as informed by the results from the global-level analysis.

Results

None of the subjects reported significant adverse side effects. One subject was excluded because EMG analyses failed to detect MEPs, even in the supposedly suprathreshold TMS conditions, possibly because of head motion after the fixation of the TMS coil. Two more subjects were excluded from the group analysis because they were “nonresponders” in whom M1 activity was not detected during suprathreshold TMS in the individual analysis at a liberal threshold (voxel-level P < 0.01 uncorrected). Hence, the results were based on data from 16 subjects. The range of stimulus intensities was between 30% and 95% of the machine output for 9 subjects and between 30% and 110% for 7 subjects.

The SSS combined with TMS yielded functional images of satisfactory quality even in scans performed just after TMS delivery (Fig. 1B). The image distortion and signal drop around the TMS coil were minimal.

Motor-Evoked Potentials

Consistent with previous literature (Ridding and Rothwell 1997), the size of EMG responses became larger as the intensity of TMS was increased during simultaneous TMS-fMRI measurement (Fig. 2A). Figure 2B illustrates the relationship between the physical intensity, expressed as a percentage of machine output, and the physiological intensity, expressed as percentage of rMT determined by standard MEP measurement within the scanner. The mean rMT across all subjects was 85.4% (SD = 13.5) of the default machine output.

Figure 2.

(A) Intensity-dependent changes in MEPs recorded from a representative subject in the MRI environment. (B) The relationship between physical intensity of stimulation relative to the default maximum machine output (MO) and physiological intensity of stimulation relative to the rMT is shown. The data are averaged across 16 subjects while the data for 100% MO are from 7 subjects only.

Figure 2.

(A) Intensity-dependent changes in MEPs recorded from a representative subject in the MRI environment. (B) The relationship between physical intensity of stimulation relative to the default maximum machine output (MO) and physiological intensity of stimulation relative to the rMT is shown. The data are averaged across 16 subjects while the data for 100% MO are from 7 subjects only.

Group-Level Statistical Parametric Mapping

The overall effects of the suprathreshold M1 stimulation on brain activity are summarized as group-level SPM reflecting the contrast between stimulation with the maximum intensity (95/110% of the default machine output depending on availability) and that with the minimal intensity (30% of the default machine output) (Table 1 and Fig. 3). The regions showing significant activity included the left M1 (stimulated site), the left S1, bilateral dorsal lateral premotor areas (PMd), the ventral part of the SMA (SMAv) extending into the caudal cingulate zone (CCZ), the rostral cingulate zone (RCZ), second somatosensory area (S2), supplementary somatosensory areas (SSA), the middle temporal gyrus, insula, thalamus, putamen, and right anteromedial cerebellum. The left-lateralized activity of the M1 and S1 centering on the “hand-knob” of the precentral gyrus and the right-lateralized anteromedial cerebellar activity were consistent with activation of the somatotopic motor/somatosensory representation of the right hand. The spatial localization of the present SMAv/CCZ regions also corresponded to the hand representations of the SMAv and CCZ (Hanakawa et al. 2008). Motor and somatosensory areas with poorer somatotopy, such as PMd and S2, were activated bilaterally. Additionally, there was activity in the auditory and cognitive/affective regions, which included the bilateral auditory cortices, hippocampus, prefrontal cortex, posterior cingulate cortex, and middle temporal cortex. The right M1 contralateral to the stimulated side did not show significant activity, but revealed a non-negligible tendency toward increases in hemodynamic responses (x = 34, y = −32, z = 54; T = 3.99) time-locked to the suprathreshold TMS.

Table 1

Results from the group-level statistical parametric mapping analysis: activity associated with suprathreshold TMS to the left primary motor cortex

Activity clusters (functional anatomy)
 
Volume (mm3Coordinates
 
T-value 
x y Z 
R inferior frontal gyrus (Area 44/45) 784 60 20 26 10.96 
R parietal operculum (OP1/OP4) 11 176 60 −20 16 10.05 
 R auditory cortex  60 −10 6.48 
 R postcentral gyrus (S1, Area 2)  62 −18 42 5.57 
 R postcentral gyrus (S1, Area 1)  44 −36 62 4.80 
L auditory cortex 16 584 −48 −20 8.81 
 L parietal operculum (OP1/OP4))  −58 −24 16 8.62 
 L insula  −38 −14 7.54 
 L putamen  −30 −10 4.97 
L middle temporal gyrus 3808 −60 −62 8.79 
R insula 8960 −40 −52 58 7.65 
 R putamen  24 18 −8 6.34 
L postcentral gyrus (S1, Area 2) 3168 −44 −36 50 8.50 
 L precentral gyrus (M1, Area 4)  −36 −24 52 7.60 
L posterior cingulate cortex 3608 −4 −40 26 7.47 
R anteromedial cerebellum 6176 16 −52 −28 7.28 
R middle cingulate cortex (RCZ) 5128 22 34 7.00 
10 L superior parietal lobule (S1, Area 1) 4712 −24 −40 72 7.47 
 L precentral gyrus (M1/PMd, Area4/6)  −40 −24 68 5.63 
11 R precuneus 416 18 −72 44 6.42 
12 R thalamus 472 12 −8 6.38 
13 Cerebellar vermis 776 −4 −46 −24 6.09 
14 R hippocampus 1040 18 −24 −10 6.82 
15 Superior frontal gyrus (SMAv, Area 6) 1304 −12 50 5.80 
 L middle cingulate cortex (SSA)  −14 −38 50 5.60 
 R paracentral lobule (SSA, Area 3)  16 −32 52 5.22 
16 R middle temporal gyrus 1096 60 −54 5.23 
17 R superior frontal gyrus (SMAd, Area 6) 168 −18 70 5.05 
18 L superior frontal gyrus (PMd, Area 6) 280 −26 −6 66 4.91 
19 R middle frontal gyrus (PMd) 536 38 58 4.69 
20 L thalamus 200 −10 −14 4.72 
21 R precentral gyrus (PMd, Area 6) 184 28 −22 72 4.60 
Activity clusters (functional anatomy)
 
Volume (mm3Coordinates
 
T-value 
x y Z 
R inferior frontal gyrus (Area 44/45) 784 60 20 26 10.96 
R parietal operculum (OP1/OP4) 11 176 60 −20 16 10.05 
 R auditory cortex  60 −10 6.48 
 R postcentral gyrus (S1, Area 2)  62 −18 42 5.57 
 R postcentral gyrus (S1, Area 1)  44 −36 62 4.80 
L auditory cortex 16 584 −48 −20 8.81 
 L parietal operculum (OP1/OP4))  −58 −24 16 8.62 
 L insula  −38 −14 7.54 
 L putamen  −30 −10 4.97 
L middle temporal gyrus 3808 −60 −62 8.79 
R insula 8960 −40 −52 58 7.65 
 R putamen  24 18 −8 6.34 
L postcentral gyrus (S1, Area 2) 3168 −44 −36 50 8.50 
 L precentral gyrus (M1, Area 4)  −36 −24 52 7.60 
L posterior cingulate cortex 3608 −4 −40 26 7.47 
R anteromedial cerebellum 6176 16 −52 −28 7.28 
R middle cingulate cortex (RCZ) 5128 22 34 7.00 
10 L superior parietal lobule (S1, Area 1) 4712 −24 −40 72 7.47 
 L precentral gyrus (M1/PMd, Area4/6)  −40 −24 68 5.63 
11 R precuneus 416 18 −72 44 6.42 
12 R thalamus 472 12 −8 6.38 
13 Cerebellar vermis 776 −4 −46 −24 6.09 
14 R hippocampus 1040 18 −24 −10 6.82 
15 Superior frontal gyrus (SMAv, Area 6) 1304 −12 50 5.80 
 L middle cingulate cortex (SSA)  −14 −38 50 5.60 
 R paracentral lobule (SSA, Area 3)  16 −32 52 5.22 
16 R middle temporal gyrus 1096 60 −54 5.23 
17 R superior frontal gyrus (SMAd, Area 6) 168 −18 70 5.05 
18 L superior frontal gyrus (PMd, Area 6) 280 −26 −6 66 4.91 
19 R middle frontal gyrus (PMd) 536 38 58 4.69 
20 L thalamus 200 −10 −14 4.72 
21 R precentral gyrus (PMd, Area 6) 184 28 −22 72 4.60 
Figure 3.

Group-level statistical parametric mapping showing the categorical comparison of hemodynamic changes between the maximal stimulation condition at the suprathreshold level (95–110% of the default machine output) and the control stimulation condition (30% of the default machine output).

Figure 3.

Group-level statistical parametric mapping showing the categorical comparison of hemodynamic changes between the maximal stimulation condition at the suprathreshold level (95–110% of the default machine output) and the control stimulation condition (30% of the default machine output).

In comparisons between the maximum and the minimum intensity conditions, there was no activity significantly greater during the minimum stimulation than during the maximum stimulation. This meant that there was no significant deactivation associated with single TMS pulses during the suprathreshold condition.

Evoked Hemodynamic Responses as a Function of Physical TMS Intensity

The directly stimulated site, the left M1, showed monotonous increases in brain activity (relative to the baseline) as a function of the physical TMS intensity, represented as a percentage of the default machine output (Fig. 4). When the physical intensity of TMS was treated as a within-subject variable for rmANOVA, significant effects of TMS intensity were evident in the left M1 (P < 0.001). In the pair-wise comparison of brain activity across different levels of stimulus intensity, M1 activity was significantly increased only in the suprathreshold conditions (90% and 100% of the default machine output) as compared with the 30% machine-output bin.

Figure 4.

Stimulus–response profiles in the directly stimulated (left M1) and remote areas with regard to the physical intensity (percentage of the default machine output) of TMS stimulation. An error bar represents the standard error of the mean. All regions showed a significant effect of stimulus intensity as shown by rmANOVA. Activity in each bin was compared with that in the 30% bin (†P < 0.05, ‡P < 0.01, #P < 0.001). M1 = primary motor cortex, S1 = primary somatosensory cortex, SMAv/CCZ = ventral supplementary motor area/caudal cingulate motor zone, S2 = second somatosensory area, RCZ = rostral cingulate zone.

Figure 4.

Stimulus–response profiles in the directly stimulated (left M1) and remote areas with regard to the physical intensity (percentage of the default machine output) of TMS stimulation. An error bar represents the standard error of the mean. All regions showed a significant effect of stimulus intensity as shown by rmANOVA. Activity in each bin was compared with that in the 30% bin (†P < 0.05, ‡P < 0.01, #P < 0.001). M1 = primary motor cortex, S1 = primary somatosensory cortex, SMAv/CCZ = ventral supplementary motor area/caudal cingulate motor zone, S2 = second somatosensory area, RCZ = rostral cingulate zone.

The effects of stimulus intensity on brain activity were significant in all sampled regions of a priori interest (left PMd P = 0.001, right PMd P = 0.001, left S1 P < 0.001, SMAv/CCZ P = 0.002, right cerebellum P < 0.001, RCZ P = 0.013, left S2 P = 0.002, right S2 P = 0.007, left thalamus P = 0.012, left auditory cortex P < 0.001, right auditory cortex P < 0.001; all by rmANOVA with nonsphericity correction) (Fig. 4). In the right PMd and SMAv/CCZ, however, brain activity responded to the subthreshold stimuli as reported previously with rTMS (Bestmann et al. 2003, 2004; Fox et al. 2006). Activity sampled from the left PMd also seemed to be increased at the levels below the motor threshold. However, this activity increase did not reach statistical significance as compared with activity in the 30% bin because of large variance of the data. Moreover, the left S1 and right cerebellum showed an almost linear pattern of intensity-dependent modulation of brain activity, which seemed to be gradually increased in response to the subthreshold stimulation. A common characteristic of the stimulus–response profile in the motor network, including both the directly stimulated and remote areas, was almost no activity in the 30% machine-output bin.

The RCZ showed a stimulus–response profile resembling that of the motor network in part, but its activity was mildly increased in response to TMS with the minimal intensity. Similarly, the nonprimary somatosensory areas (bilateral S2) showed substantial activity in the 30% bin and prominent increases in activity in the suprathreshold conditions. It seemed that activity in the S2 and RCZ, most typically in the right S2, primarily represented the states of stimulation (suprathreshold TMS that evoked movement or subthreshold activity that did not). The left thalamus showed a similar pattern. The auditory cortices showed robust activity even in the 30% bin. In addition, activity in the auditory cortices was very clearly modulated by physical TMS intensity in a parametric fashion.

Evoked Hemodynamic Responses as a Function of Physiological TMS Intensity

The stimulus–response profile was analyzed by reformatting the data with regard to the rMT in each individual (Fig. 5). First, a global pattern of stimulus–response profiles was investigated by including all levels of TMS intensity relative to the rMT (Fig. 5A). In this global-level analysis, many areas showed a mixture of linear and nonlinear components in the stimulus–response profile. A nonlinear feature was prominent in the left M1 and S2, as evident in the curve fitting with a sigmoid function (Boltzmann function). In these areas, the brain responses showed a sharp increase around the rMT. Linearity of the stimulus–response profile was rather pronounced in the SMAv/CCZ and the right cerebellum, but fair fitting with a sigmoid function, suggesting that a nonlinear component around the rMT also existed in these areas. Notably, however, the hemodynamic responses in the SMAv/CCZ and cerebellum appeared to be almost linearly increased below the subthreshold level of intensity. Fitting with a sigmoid function apparently failed to capture these changes occurring at the subthreshold stimulation level. Therefore, these subthreshold-level linear components were separately assessed, as described later in the subthreshold-level analysis. In contrast to the above motor/somatosensory areas, the nonlinear component was less evident in the bilateral auditory cortices. To verify the adequacy of applying sigmoid fitting to the global-level analysis, the proportion of data variance explained by fitting was compared between sigmoid fitting and linear fitting. For this purpose, each fitting was performed by using all of the stimulation conditions in each individual. The resultant coefficient of determination was used as a summary variable of goodness of fitting and fed into the group-level nonparametric statistics (Wilcoxon signed-rank test). It was shown that the sigmoid fitting better explained the data than the linear fitting in the M1 (P = 0.017, 46.4% of data variance explained by sigmoid fitting and 32.2% by linear fitting), SMA/CCZ (P = 0.002, 36.9% by sigmoid fitting and 25.1% by linear fitting), and S2 (P = 0.002, 35.9% by sigmoid fitting and 18.8% by linear fitting). In the right cerebellum, however, no difference was found (P = 0.88) between sigmoid (26.4%) and linear fittings (25.3%). In the left auditory cortex, although data variance tended to be better explained by linear fitting (35.5%) than sigmoid fitting (25.0%), the difference did not reach statistical significance (P = 0.08).

Figure 5.

(A) Global-level analysis of stimulus–response profiles as fitted with a sigmoid function. Stimulus intensity is expressed as the physiological intensity (% of rMT) of TMS stimulation. All the sub- and suprathreshold stimulus conditions were taken into account. Nonlinearity is evident at around 100% of the rMT in the directly stimulated left primary motor cortex (M1) as well as in the SMAv/CCZ, and the right second somatosensory area (S2). The left auditory cortex showed an almost linear stimulus–response profile. The gray dots represent a single data point from each stimulus-intensity condition from each subject. (B) Subthreshold-level analysis of physiological intensity-response profiles as fitted with a linear function. All areas except the right S2 showed a mild yet significant increase in activity as a linear function of stimulus intensity. The gray dots represent a single data point from each stimulus-intensity condition from each subject.

Figure 5.

(A) Global-level analysis of stimulus–response profiles as fitted with a sigmoid function. Stimulus intensity is expressed as the physiological intensity (% of rMT) of TMS stimulation. All the sub- and suprathreshold stimulus conditions were taken into account. Nonlinearity is evident at around 100% of the rMT in the directly stimulated left primary motor cortex (M1) as well as in the SMAv/CCZ, and the right second somatosensory area (S2). The left auditory cortex showed an almost linear stimulus–response profile. The gray dots represent a single data point from each stimulus-intensity condition from each subject. (B) Subthreshold-level analysis of physiological intensity-response profiles as fitted with a linear function. All areas except the right S2 showed a mild yet significant increase in activity as a linear function of stimulus intensity. The gray dots represent a single data point from each stimulus-intensity condition from each subject.

Finally in the subthreshold-level analysis, the evaluation of the stimulus–response profile was limited to the subthreshold condition (intensity < 80% of rMT). Because this analysis was meant to assess linear increases in the hemodynamic signals at the subthreshold level, fitting was performed only with a linear function. In response to subthreshold stimulation, many areas showed a linear increase in brain activity as a function of physiological intensity (Fig. 5B). This linear increase in hemodynamic signals at subthreshold-level stimulation was statistically significant in the SMA/CCZ, right cerebellum, auditory cortex, and, importantly, also in the directly stimulated left M1. The only exception was the right S2.

Discussion

The present results have revealed both linear and nonlinear aspects in the stimulus–response relationship between the intensity of single TMS pulses and evoked hemodynamic responses in the directly stimulated and remote motor/somatosensory network. These results not only replicated previous findings from a TMS-EEG study (Komssi et al. 2004), but also identified multiple sources of linear and nonlinear neural responses in the motor/somatosensory network.

A couple of imaging studies have addressed the issue of a TMS intensity-hemodynamic response profile, using rTMS (1–4 Hz) with a train of at least 10 pulses (Bestmann et al. 2003; Speer et al. 2003; Fox et al. 2006). It should be remembered that rTMS modulates cortical excitability in a complicated manner as an interaction among stimulus frequency, intensity, and length of pulse trains. Indeed, a few TMS-fMRI/PET studies examining the effects of the length of pulse trains or stimulus frequency have produced conflicting results. When brain activity was assessed as a function of the length of pulse trains, a TMS-fMRI study reported a linear increase in hemodynamic signals (frequency 1 Hz, intensity 120% rMT) (Bohning et al. 2003), whereas a TMS-PET study found a linear decrease in blood flow (frequency 1 Hz, subthreshold intensity) (Paus et al. 1998). Another TMS-PET study found a linear increase in blood flow as a function of the stimulus frequency (1–5 Hz) of a continuous rTMS with subthreshold intensity (Siebner et al. 2001). The discrepancy can be partly explained by differential effects of rTMS on cortical excitability, depending on whether they are delivered continuously or intermittently (Huang et al. 2005).

The present study employed single TMS pulses at a frequency <0.2 Hz, which less likely had carry-over effects on cortical excitability (Chen et al. 1997). Our study covered the widest range of intensities ever studied using the concurrent TMS-fMRI/PET method, as it seemed possible that the range of intensities used in previous studies might not be sufficiently wide to delineate the full picture of the stimulus–response profile. The present study should thus reveal one of the most fundamental forms of the stimulus–response profile, perhaps along with a study examining EEG responses during single-pulse TMS (Komssi et al. 2004). A common finding between that study and the present one was that of both linear and nonlinear components in the relationship between the intensities of single TMS pulses and evoked brain responses. The present study indicates for the first time that this effect can be approximated by a linear function of stimulus intensity at the subthreshold level, but nonlinear components emerge at the levels of stimulus intensity around the motor threshold. These findings implicate that even single TMS pulses may influence neural activity in the remote motor/somatosensory areas in both linear and nonlinear fashions, depending on the stimulus intensity. These lines of evidence should help refine interpretation of the effects of single TMS pulses on behavior and MEP.

Stimulus–Response Profile in the Motor and Sensory Networks

The cortical motor areas, including the left M1, SMAv/CCZ, and bilateral PMd, commonly demonstrated an almost monotonous increase in activity as a function of physical TMS intensity. Also in common, these directly stimulated and remote motor areas showed almost no activity during TMS with the lowest intensity studied here (30% of the default machine output). The directly stimulated M1 revealed significant activity only in the suprathreshold conditions relative to the minimal intensity condition. However, remote motor areas except for the left PMd showed significant activity, even during subthreshold stimulation, relative to the minimal intensity condition. Previous TMS-fMRI/PET studies with M1 stimulation have already pointed out that subthreshold TMS activates remote motor regions, but not the directly stimulated M1 (Bestmann et al. 2003, 2004; Fox et al. 2006). Activity in the M1 was only detected during suprathreshold TMS (Baudewig et al. 2001; Bestmann et al. 2003, 2004; Fox et al. 2006). These observations have led the researchers to hypothesize that activity in the stimulated M1 mainly reflects afferent sensory information from the twitched muscles. However, the sensory afferent theory may not be the sole explanation for the behavior of the directly stimulated M1 activity. A parametric-design PET study with 1-Hz rTMS showed increases in M1 activity during subthreshold stimulation (80% or 90% rMT), although the activities were clearly smaller than those during suprathreshold stimulation (Speer et al. 2003). During median nerve stimulation inducing clear twitching of hand muscles, many imaging studies only reported activity in the contralateral S1 (see e.g., Del Gratta et al. 2000; Ferretti et al. 2007), although a limited number of studies reported M1 activity in addition to much larger S1 activity (Spiegel et al. 1999). Finally, a similar phenomenon (activity in the directly stimulated region only during suprathreshold stimulation) is also observed in the prefrontal cortex (Nahas et al. 2001) and the PMd (Bestmann et al. 2005), where TMS induces no movement.

With rTMS protocols, an increase in cerebellar activity contralateral to the stimulated M1 was previously reported in response to suprathreshold- and threshold-level stimulation (Speer et al. 2003; Bestmann et al. 2004), whereas a recent study found linear decreases in cerebellar activity as a function of stimulus intensity (Fox et al. 2006). This discrepancy might result from a complex interaction among the stimulus parameters of rTMS. In the present single-pulse TMS-fMRI study, the stimulus–response profile of the right anteromedial cerebellum resembled that of the M1. It is tempting to assume that the cerebellar activity reflects not only the afferents from the muscles, but also remote effects of M1 stimulation, because there was a parametric increase in activity even in the subthreshold condition. The cerebellar remote activity in the suprathreshold condition could partly be ascribed to the “efference copy” of the motor commands from the M1 to the spinal cord.

The S2 in either hemisphere revealed mild activity during subthreshold stimulation, including stimulation with 30% of the machine output and much enhanced activity during suprathreshold stimulation. The enhanced suprathreshold activity was consistent with the roles of the S2 in analyzing sensory afferents. The similar pattern of stimulus–response profile was observed in the RCZ and the left thalamus, which could partially reflect cognitive/affective components resulting from TMS stimulation. This interpretation is consistent with the widespread activity in the cognitive/affective regions including the prefrontal cortex and the hippocampi. Meanwhile, whereas most of the sampled areas showed parametric activity increases in the subthreshold stimulation, the right S2 did not (Fig. 5B). It appeared that the right S2 coded the state of stimulation: stimulation without muscle twitches or stimulation with muscle twitches.

Previous concurrent TMS-fMRI/PET studies often reported activity in the auditory cortex (Bestmann et al. 2004, 2005; Bohning et al. 1998, 2000; Siebner et al. 1998). Click sounds associated with TMS pulses were especially pronounced because of the much stronger force exerted on the TMS coil in MRI environments compared with non-MRI environments. The vibration of the coil and the loudness of click sounds should be a function of TMS intensity, and activity in the auditory cortices was almost linearly increased in both the physical and physiological intensity analyses. Activity in the middle temporal gyrus is consistent with previous findings during suprathreshold TMS to the M1 (Bestmann et al. 2004) or to the PMd (Bestmann et al. 2005), but its significance is not clear.

Because of the reasons explained in the method section, interpretation of activity sampled from the left PMd and S1 is only speculative. The interaction between the remote and direct effects of TMS may partly explain the variability of the data in the left PMd. Although the left S1, which should process somatosensory information caused by muscle twitch, was highly activated in the suprathreshold conditions as expected, it also demonstrated significant activation in the subthreshold conditions. Subthreshold-level activity in the left S1 could primarily reflect remote effects as it closely resembled the pattern observed in the SMA/CCZ and the cerebellum. The stimulus–response profile in the right M1, which may show remote activity through the transcallosal connection with the stimulated M1, also seems interesting. This analysis was not reported as the right M1 activity did not reach significance in the group-level SPM analysis, although it was moderately increased during suprathreshold single TMS pulses with the present stimulus protocol. This finding can be contrasted to previous reports of decreased contralateral M1 activity with rTMS protocols (Baudewig et al. 2001; Bestmann et al. 2003, 2004; Fox et al. 2006), suggesting the dependency of TMS-induced hemodynamic changes on stimulus protocols.

Possible Sources of Linear and Nonlinear Aspects of Stimulus–Response Property

The dose–response profile of the M1 activity clearly exhibited a nonlinear component, which was a sharp rise in activity starting around the rMT. As a similar activity rise was observed in the S2, it is likely that the nonlinear component at least partly reflected sensory afferents. However, a milder degree of nonlinearity was also observed in the SMAv/CCZ, in which activity started to rise gradually from the subthreshold level. This difference in the stimulus–response profile between the directly stimulated and remote motor areas is consistent with previous observations that the remote areas revealed activity with weaker stimulation than did the directly stimulated M1 (Bestmann et al. 2003, 2004).

Now, the question is what the sources of the nonlinearity in the directly stimulated and remote areas might be. Another point requiring attention is that both the directly stimulated and remote areas revealed a mild increase in activity as a linear function of subthreshold-level intensity (see Fig. 5B). The profile in the right S2 (that is, no parametric changes during subthreshold stimulation) does not support the idea that such stimulus–response profiles during subthreshold stimulation can solely be explained by changes in attention levels related to larger clicks. It is hence probable that at least some of the aforementioned linear activity increases during subthreshold stimulation truly reflect direct and remote effects of M1. This idea is consistent with recent findings from an invasive cortico-cortical evoked potential study in which subthreshold electrical stimulation to the M1 induced responses in remote motor areas, including the SMA (Matsumoto et al. 2007).

A following idea could explain the observed behavior of brain activity in the directly stimulated and remote motor areas in the present study. Single TMS pulses induce a sequence of excitatory and inhibitory neuronal processes locally in the stimulated region in response to a wide range of stimulus intensities, including subthreshold levels (Moliadze et al. 2003). It may be that, below the threshold, the excitatory activity does not surpass the inhibitory activity to activate long projecting neurons to the spinal cord. Around the threshold, the excitatory processes would become predominant and start to depolarize a larger number of long projecting neurons. This recruitment process is conceivably nonlinear, provided that the depolarizing threshold of each long projecting neuron is distributed normally around the rMT. This hypothesis would expect induction of both the direct and remote activities in a nonlinear fashion. Furthermore, with hemodynamic measurement, activity can be detected more easily in the remote areas than the directly stimulated area because 1) hemodynamic signals largely reflect neural inputs to a particular region in addition to local spiking activity (Logothetis et al. 2001), and 2) remote activity, perhaps induced by release of neurotransmitter of transcortical neurons, may consume more metabolic resources than direct activity because energy for depolarizing neurons is provided externally in the direct area, at least in part. These considerations may account for the nonlinear component in the remote and directly stimulated areas from the viewpoint of not only sensory afferents but also TMS-induced neuronal changes.

Methodological Consideration

The present study has, for the first time, applied the SSS scheme (Anami et al. 2003) to a concurrent TMS-fMRI study for monitoring and recording of MEPs. Because the feasibility of concurrent TMS-fMRI was first demonstrated by Bohning and colleagues (Bohning et al. 1998), this innovative multidisciplinary mapping technique has provided new insight into direct and remote actions of TMS. Online measurement of EMGs may enable further technical advances in combined TMS-fMRI studies (Bestmann et al. 2004, 2005). First, it is ideal to determine formal rMT in a real experimental environment. The fact that rMT was relatively high in the present setup can largely be explained by the long cable between the TMS coil and the stimulator plus difficulty in positioning the TMS coil within the MRI head coil. In this regard, it is obvious that applying an rMT determined outside the MRI is not ideal. Second, constant MEPs should guarantee stable positioning of the TMS coil relative to the head during the experiment. Finally, unintended movement or muscle activity, which would influence both MEP and hemodynamic signals, can be monitored.

It could be a matter for debate whether we can apply the same hemodynamic response function to directly stimulated and remote regions. A previous concurrent TMS-fMRI study addressed this question and found a standard hemodynamic response to a single TMS pulse consistently across the stimulated M1 and auditory cortices (Bohning et al. 2000). In accordance with this finding, we used the standard hemodynamic response function for the SPM analysis and were able to capture reasonable activity from all regions of interest. Then, a VOI analysis, rather than a voxel-wise parametric approach, was employed as both linear and nonlinear components of the stimulus–response profile were of a priori interest.

In conclusion, the present study demonstrated for the first time a detailed stimulus–response profile in the neural network consisting of directly stimulated and remote activations induced during single TMS pulses, by combining MEP and hemodynamic measurements at 3 Tesla. Single TMS pulses evoked activity in the motor/somatosensory network and other areas related to detection of TMS delivery. The stimulus-intensity profiles revealed both linear and nonlinear components. The nonlinear increase in brain activity in the directly stimulated M1 may be induced by the nonlinear recruitment of neurons in addition to sensory afferents.

Funding

Grant-in-Aid for Scientific Research on Priority Areas (Integrative Brain Research 20019041) to T.H.; (20019023) to T.M.; (20020013) to H.F.; Emergence of Adaptive Motor Function through Interaction among the Body, Brain and Environment (20033030) to T.H.; Strategic Research Program for Brain Sciences (SRPBS) to T.M. from the Ministry of Education, Science, Sports, Culture, and Technology of Japan; Grant-in-Aid for Scientific Research (C) (18500239) to T.M. from the Japan Society for the Promotion of Science; a Grant-in-Aid from the Takeda Science Foundation (2007) to T.H.; and a research grant (2007) from Neurocreative Lab to T.M.

Conflict of Interest: None declared.

References

Allen
EA
Pasley
BN
Duong
T
Freeman
RD
Transcranial magnetic stimulation elicits coupled neural and hemodynamic consequences
Science.
 , 
2007
, vol. 
317
 (pg. 
1918
-
1921
)
Anami
K
Mori
T
Tanaka
F
Kawagoe
Y
Okamoto
J
Yarita
M
Ohnishi
T
Yumoto
M
Matsuda
H
Saitoh
O
Stepping stone sampling for retrieving artifact-free electroencephalogram during functional magnetic resonance imaging
NeuroImage.
 , 
2003
, vol. 
19
 (pg. 
281
-
295
)
Baudewig
J
Siebner
HR
Bestmann
S
Tergau
F
Tings
T
Paulus
W
Frahm
J
Functional MRI of cortical activations induced by transcranial magnetic stimulation (TMS)
Neuroreport.
 , 
2001
, vol. 
12
 (pg. 
3543
-
3548
)
Bestmann
S
Baudewig
J
Siebner
HR
Rothwell
JC
Frahm
J
Subthreshold high-frequency TMS of human primary motor cortex modulates interconnected frontal motor areas as detected by interleaved fMRI-TMS
NeuroImage.
 , 
2003
, vol. 
20
 (pg. 
1685
-
1696
)
Bestmann
S
Baudewig
J
Siebner
HR
Rothwell
JC
Frahm
J
Functional MRI of the immediate impact of transcranial magnetic stimulation on cortical and subcortical motor circuits
Eur J Neurosci.
 , 
2004
, vol. 
19
 (pg. 
1950
-
1962
)
Bestmann
S
Baudewig
J
Siebner
HR
Rothwell
JC
Frahm
J
BOLD MRI responses to repetitive TMS over human dorsal premotor cortex
NeuroImage.
 , 
2005
, vol. 
28
 (pg. 
22
-
29
)
Bohning
DE
Shastri
A
Lomarev
MP
Lorberbaum
JP
Nahas
Z
George
MS
BOLD-fMRI response vs. transcranial magnetic stimulation (TMS) pulse-train length: testing for linearity
J Magn Reson Imaging.
 , 
2003
, vol. 
17
 (pg. 
279
-
290
)
Bohning
DE
Shastri
A
McConnell
KA
Nahas
Z
Lorberbaum
JP
Roberts
DR
Teneback
C
Vincent
DJ
George
MS
A combined TMS/fMRI study of intensity-dependent TMS over motor cortex
Biol Psychiatry.
 , 
1999
, vol. 
45
 (pg. 
385
-
394
)
Bohning
DE
Shastri
A
Nahas
Z
Lorberbaum
JP
Andersen
SW
Dannels
WR
Haxthausen
EU
Vincent
DJ
George
MS
Echoplanar BOLD fMRI of brain activation induced by concurrent transcranial magnetic stimulation
Invest Radiol.
 , 
1998
, vol. 
33
 (pg. 
336
-
340
)
Bohning
DE
Shastri
A
Wassermann
EM
Ziemann
U
Lorberbaum
JP
Nahas
Z
Lomarev
MP
George
MS
BOLD-f MRI response to single-pulse transcranial magnetic stimulation (TMS)
J Magn Reson Imaging.
 , 
2000
, vol. 
11
 (pg. 
569
-
574
)
Chen
R
Classen
J
Gerloff
C
Celnik
P
Wassermann
EM
Hallett
M
Cohen
LG
Depression of motor cortex excitability by low-frequency transcranial magnetic stimulation
Neurology.
 , 
1997
, vol. 
48
 (pg. 
1398
-
1403
)
Del Gratta
C
Della Penna
S
Tartaro
A
Ferretti
A
Torquati
K
Bonomo
L
Romani
GL
Rossini
PM
Topographic organization of the human primary and secondary somatosensory areas: an fMRI study
Neuroreport.
 , 
2000
, vol. 
11
 (pg. 
2035
-
2043
)
Eickhoff
SB
Stephan
KE
Mohlberg
H
Grefkes
C
Fink
GR
Amunts
K
Zilles
K
A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data
NeuroImage.
 , 
2005
, vol. 
25
 (pg. 
1325
-
1335
)
Ferretti
A
Babiloni
C
Arienzo
D
Del Gratta
C
Rossini
PM
Tartaro
A
Romani
GL
Cortical brain responses during passive nonpainful median nerve stimulation at low frequencies (0.5-4 Hz): an fMRI study
Hum Brain Mapp.
 , 
2007
, vol. 
28
 (pg. 
645
-
653
)
Fox
P
Ingham
R
George
MS
Mayberg
H
Ingham
J
Roby
J
Martin
C
Jerabek
P
Imaging human intra-cerebral connectivity by PET during TMS
Neuroreport.
 , 
1997
, vol. 
8
 (pg. 
2787
-
2791
)
Fox
PT
Narayana
S
Tandon
N
Fox
SP
Sandoval
H
Kochunov
P
Capaday
C
Lancaster
JL
Intensity modulation of TMS-induced cortical excitation: primary motor cortex
Hum Brain Mapp.
 , 
2006
, vol. 
27
 (pg. 
478
-
487
)
Hanakawa
T
Dimyan
MA
Hallett
M
The representation of blinking movement in cingulate motor areas: a functional magnetic resonance imaging study
Cereb Cortex.
 , 
2008
, vol. 
18
 (pg. 
930
-
937
)
Hayashi
T
Ohnishi
T
Okabe
S
Teramoto
N
Nonaka
Y
Watabe
H
Imabayashi
E
Ohta
Y
Jino
H
Ejima
N
, et al.  . 
Long-term effect of motor cortical repetitive transcranial magnetic stimulation
Ann Neurol.
 , 
2004
, vol. 
56
 (pg. 
77
-
85
)
Huang
YZ
Edwards
MJ
Rounis
E
Bhatia
KP
Rothwell
JC
Theta burst stimulation of the human motor cortex
Neuron.
 , 
2005
, vol. 
45
 (pg. 
201
-
206
)
Komssi
S
Kahkonen
S
Ilmoniemi
RJ
The effect of stimulus intensity on brain responses evoked by transcranial magnetic stimulation
Hum Brain Mapp.
 , 
2004
, vol. 
21
 (pg. 
154
-
164
)
Logothetis
NK
Pauls
J
Augath
M
Trinath
T
Oeltermann
A
Neurophysiological investigation of the basis of the fMRI signal
Nature.
 , 
2001
, vol. 
412
 (pg. 
150
-
157
)
Matsumoto
R
Nair
DR
LaPresto
E
Bingaman
W
Shibasaki
H
Luders
HO
Functional connectivity in human cortical motor system: a cortico-cortical evoked potential study
Brain.
 , 
2007
, vol. 
130
 (pg. 
181
-
197
)
Moliadze
V
Zhao
Y
Eysel
U
Funke
K
Effect of transcranial magnetic stimulation on single-unit activity in the cat primary visual cortex
J Physiol.
 , 
2003
, vol. 
553
 (pg. 
665
-
679
)
Nahas
Z
Lomarev
M
Roberts
DR
Shastri
A
Lorberbaum
JP
Teneback
C
McConnell
K
Vincent
DJ
Li
X
George
MS
Bohning
DE
Unilateral left prefrontal transcranial magnetic stimulation (TMS) produces intensity-dependent bilateral effects as measured by interleaved BOLD fMRI
Biol Psychiatry.
 , 
2001
, vol. 
50
 (pg. 
712
-
720
)
Okabe
S
Hanajima
R
Ohnishi
T
Nishikawa
M
Imabayashi
E
Takano
H
Kawachi
T
Matsuda
H
Shiio
Y
Iwata
NK
, et al.  . 
Functional connectivity revealed by single-photon emission computed tomography (SPECT) during repetitive transcranial magnetic stimulation (rTMS) of the motor cortex
Clin Neurophysiol.
 , 
2003
, vol. 
114
 (pg. 
450
-
457
)
Paus
T
Jech
R
Thompson
CJ
Comeau
R
Peters
T
Evans
AC
Transcranial magnetic stimulation during positron emission tomography: a new method for studying connectivity of the human cerebral cortex
J Neurosci.
 , 
1997
, vol. 
17
 (pg. 
3178
-
3184
)
Paus
T
Jech
R
Thompson
CJ
Comeau
R
Peters
T
Evans
AC
Dose-dependent reduction of cerebral blood flow during rapid-rate transcranial magnetic stimulation of the human sensorimotor cortex
J Neurophysiol.
 , 
1998
, vol. 
79
 (pg. 
1102
-
1107
)
Penfield
W
Boldrey
E
Somatic motor and sensory representation in the cerebral cortex of man as studied by electrical stimulation
Brain.
 , 
1937
, vol. 
60
 (pg. 
389
-
443
)
Ridding
MC
Rothwell
JC
Stimulus/response curves as a method of measuring motor cortical excitability in man
Electroencephalogr Clin Neurophysiol.
 , 
1997
, vol. 
105
 (pg. 
340
-
344
)
Rossini
PM
Barker
AT
Berardelli
A
Caramia
MD
Caruso
G
Cracco
RQ
Dimitrijevic
MR
Hallett
M
Katayama
Y
Lucking
CH
, 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
, vol. 
91
 (pg. 
79
-
92
)
Siebner
HR
Takano
B
Peinemann
A
Schwaiger
M
Conrad
B
Drzezga
A
Continuous transcranial magnetic stimulation during positron emission tomography: a suitable tool for imaging regional excitability of the human cortex
NeuroImage.
 , 
2001
, vol. 
14
 (pg. 
883
-
890
)
Siebner
HR
Willoch
F
Peller
M
Auer
C
Boecker
H
Conrad
B
Bartenstein
P
Imaging brain activation induced by long trains of repetitive transcranial magnetic stimulation
Neuroreport.
 , 
1998
, vol. 
9
 (pg. 
943
-
948
)
Speer
AM
Willis
MW
Herscovitch
P
Daube-Witherspoon
M
Shelton
JR
Benson
BE
Post
RM
Wassermann
EM
Intensity-dependent regional cerebral blood flow during 1-Hz repetitive transcranial magnetic stimulation (rTMS) in healthy volunteers studied with H215O positron emission tomography: I. Effects of primary motor cortex rTMS
Biol Psychiatry.
 , 
2003
, vol. 
54
 (pg. 
818
-
825
)
Spiegel
J
Tintera
J
Gawehn
J
Stoeter
P
Treede
RD
Functional MRI of human primary somatosensory and motor cortex during median nerve stimulation
Clin Neurophysiol.
 , 
1999
, vol. 
110
 (pg. 
47
-
52
)