- Email: [email protected]
fMRI of the Human Motor Cortex Using a Dedicated Multichannel MR Coil
NeuroImage 150 (2017) 262–269
Contents lists available at ScienceDirect
NeuroImage journal homepage: www.elsevier.com/locate/neuroimage
High-sensitivity TMS/fMRI of the Human Motor Cortex Using a Dedicated Multichannel MR Coil
MARK
Lucia I. Navarro de Laraa,b,1, Martin Tika,b,1, Michael Woletza,b, Roberta Frass-Kriegla,b, ⁎ Ewald Mosera,b, Elmar Laistlera,b, Christian Windischbergera,b, a b
Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Währinger Guertel 18–20, A-1090 Wien, Vienna, Austria MR Center of Excellence, Medical University of Vienna, Vienna, Austria
A R T I C L E I N F O
A BS T RAC T
Keywords: TMS fMRI Coil array parallel imaging motor cortex concurrent TMS/fMRI
Purpose: To validate a novel setup for concurrent TMS/fMRI in the human motor cortex based on a dedicated, ultra-thin, multichannel receive MR coil positioned between scalp and TMS system providing greatly enhanced sensitivity compared to the standard birdcage coil setting. Methods: A combined TMS/fMRI design was applied over the primary motor cortex based on 1 Hz stimulation with stimulation levels of 80%, 90%, 100%, and 110% of the individual active motor threshold, respectively. Due to the use of a multichannel receive coil we were able to use multiband-accelerated (MB=2) EPI sequences for the acquisition of functional images. Data were analysed with SPM12 and BOLD-weighted signal intensity time courses were extracted in each subject from two local maxima (individual functional finger tapping localiser, fixed MNI coordinate of the hand knob) next to the hand area of the primary motor cortex (M1) and from the global maximum. Results: We report excellent image quality without noticeable signal dropouts or image distortions. Parameter estimates in the three peak voxels showed monotonically ascending activation levels over increasing stimulation intensities. Across all subjects, mean BOLD signal changes for 80%, 90%, 100%, 110% of the individual active motor threshold were 0.43%, 0.63%, 1.01%, 2.01% next to the individual functional finger tapping maximum, 0.73%, 0.91%, 1.34%, 2.21% next to the MNI-defined hand knob and 0.88%, 1.09%, 1.65%, 2.77% for the global maximum, respectively. Conclusion: Our results show that the new setup for concurrent TMS/fMRI experiments using a dedicated MR coil array allows for high-sensitivity fMRI particularly at the site of stimulation. Contrary to the standard birdcage approach, the results also demonstrate that the new coil can be successfully used for multibandaccelerated EPI acquisition. The gain in flexibility due to the new coil can be easily combined with neuronavigation within the MR scanner to allow for accurate targeting in TMS/fMRI experiments.
Introduction Transcranial magnetic stimulation (TMS) is a non-invasive brain stimulation method widely applied in scientific research and a promising clinical diagnostic tool (Walsh and Pascual-Leone, 2005, Rossini and Rossi, 2007, Chen et al., 2008) and treatment option (Lefaucheur et al., 2014). It allows for interacting with the function of brain regions at well-defined time points and thus, establishment of causal relationships between the region stimulated and behaviour (Silvanto and Pascual-Leone, 2012). The combination of TMS with functional magnetic resonance imaging (fMRI) has received considerable attention over the last years because it promises to provide a better
⁎
1
understanding of the mechanisms underlying TMS effects. So-called “online” approaches, where TMS stimulation is applied to subjects during fMRI acquisition, are of particular interest. Although early pioneering work already demonstrated the feasibility of this combination (Bohning et al., 1998, Bohning et al., 1999), online TMS/fMRI still poses a number of challenges. The current state-of-the-art setup for concurrent TMS/fMRI uses a large-volume head coil, usually a large birdcage coil, where the MRcompatible TMS coil is mounted inside the volume coil. This setup has been used to explore mainly the primary motor cortex as M1 stimulation results can be observed as visible finger twitches and quantified as motor evoked potentials (MEPs) using Electromyography (EMG).
Corresponding author at: Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Währinger Guertel 18–20, A-1090 Wien, Vienna, Austria. E-mail address: [email protected] (C. Windischberger). Contributed equally.
http://dx.doi.org/10.1016/j.neuroimage.2017.02.062 Received 23 June 2016; Accepted 21 February 2017 Available online 22 February 2017 1053-8119/ © 2017 Elsevier Inc. All rights reserved.
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
coils) were accurately studied demonstrating the absence of significant induction currents in the elements of the imaging coil array during TMS stimulation. Another point addressed in the publication was the difference in quality and quantity of the field intensity produced by the TMS with and without the presence of the MR coil between the TMS and a phantom. It was shown that the only effect of the MR coil presence was a loss of not more than 15% of the field intensity, without producing any distortion in TMS field distribution. In addition, the multi-channel receiver approach with the novel MR coil allowed acquisition at high temporal and spatial resolution employing parallel imaging and multi-band techniques. In the present study we aimed for demonstrating the benefits of the novel RF coil setup in concurrent TMS/fMRI experiments focused on the primary motor cortex. We acquired data with unprecedented high spatial and temporal resolution. The absence of a birdcage coil also allowed for straight-forward application of neuronavigation tools inside the scanner as no surrounding volume coil limited the visibility of retro-reflective markers. From a data acquisition perspective, we tested the use of multi-band (also referred to as simultaneous multi-slice, SMS) imaging with the new coil setup to benefit from significant scan time reduction (Auerbach et al., 2013). Based on the expected gain in sensitivity we also aimed for analysing the relationship between the TMS-induced BOLD signal change amplitudes and the TMS intensities applied at the stimulation site.
Several studies have been conducted using this setup focusing on local effects and motor network modulations. For other experiments where the observed region was not located directly under the TMS coil, a surface coil placed above the investigated region was used to increase fMRI sensitivity. This setup was successfully used for visual cortex experiments where stimulation was applied to parietal and frontal regions (Ruff et al., 2006, Ruff et al., 2008, Ruff et al., 2009). The first experiments applied TMS at a frequency of 1 Hz over the left motor cortex, with either changing stimulation intensities (Bohning et al., 1999) or pulse train lengths (Bohning et al., 2003). Extension to full brain coverage resulted in activation of subcortical areas for this particular TMS frequency (Denslow et al., 2005). It was also shown that higher stimulation frequencies are feasible via inter-slice stimulation approaches (Baudewig et al., 2001, Bestmann et al., 2003b, 2004). When applying subthreshold stimulation, significant signal changes in remote regions were detected in anatomically interconnected areas to stimulation target primary motor cortex (M1), such as supplementary motor area (SMA), cingulate motor region, and left dorsal premotor cortex (Baudewig et al., 2001, Bestmann et al., 2003b, 2004). Surprisingly, no local activation changes were observed at stimulation below motor threshold. Since pulses above but not below motor threshold led to measurable changes in BOLD (Blood oxygenation level dependent) directly at M1, it was proposed that activation in this region for supra-threshold pulses could be influenced by reafferent feedback from the muscle twitch (Baudewig et al., 2001). BOLD response to single TMS pulses over the motor cortex has been investigated using interleaved TMS/fMRI (Bohning et al., 2000, Bestmann et al., 2006, Hanakawa et al., 2009, Shitara et al., 2011). They validated the use of the hemodynamic response function for TMSinduced fMRI changes and proposed linear and non-linear models for different brain regions as response to different TMS stimulation intensities. Other studies investigated the functional interaction between dorsal premotor cortex and interconnected motor regions during different motor tasks (Bestmann et al., 2008, Moisa et al., 2012). In these experiments, TMS was applied for perturbing an on-going task in order to measure local and remote effects. For a general review of combining TMS with neuroimaging see e.g. (Siebner et al., 2009, Bestmann and Feredoes, 2013). Important efforts were made to resolve various technical problems presented by the combination of TMS and fMRI. Artefacts in echoplanar imaging can be minimized if TMS pulses are applied at least 100 ms before EPI acquisition and no TMS pulses are applied during imaging (Bestmann et al., 2003a). Leakage-current artefacts produced by the TMS coil can be minimized using a relay-diode combination (Weiskopf et al., 2009). Susceptibility effects due to the TMS coil were compensated using passive shimming in order to avoid image distortions and signal dropouts (Bungert et al., 2012). However, the most crucial challenge in these experiments is caused by the limitation in signal-to-noise ratio (SNR) due to the use of largediameter birdcage coils, resulting in poor MR sensitivity. A birdcage coil is usually limited to a single (quadrature) RF channel and achieves considerably lower signal-to-noise ratio compared to the current stateof-the-art multi-channel receive coil arrays (Wiggins et al., 2006). In addition, TMS application is severely hampered by the space limits within birdcage coils, reducing patient comfort and positioning choices. To overcome these limitations we developed a dedicated MR multi-channel receive coil array for concurrent TMS/fMRI experiments (Navarro de Lara et al., 2015) shown in Fig. 1a, which allows for very flexible TMS positioning (see Fig. 1b). There it was shown that the new setup yields more than 5-times higher SNR at a target depth of 3 cm compared to the state-of-the-art birdcage coil approach. Interactions between both systems (i.e. TMS and RF
Methods Subjects Seven young healthy subjects (all females, age: 26.1 ± 3.6 years) participated in the experiments after giving written informed consent. None of them reported any history of neuropsychiatric disease and no clinical evidence of neurological dysfunction. All subjects were consistently right-handed as assessed with the Edinburgh Handedness Inventory EHI (Oldfield, 1971). The mean handedness score on the EHI was 83 (SD 13). The study was approved by the ethics board of the Medical University of Vienna and conducted according to the Declaration of Helsinki. Experimental protocol The experiment consisted of two sessions. The manufacturer's standard 32-channel head coil (Siemens, Erlangen, Germany) was used in the first session, while in the second session data were acquired using the novel dedicated MR coil array for combined TMS/fMRI, developed by our group (Navarro de Lara et al., 2015). In the first session where data was acquired with the standard 32channel head coil, subjects were asked to perform a standard finger tapping paradigm using the right hand (nine 10 s blocks each separated by 10 s rest) and an anatomical scan was acquired for reference. The results of this session were used as neuronavigation-based target in the following TMS/fMRI session. The second session, where the new receive-array was used for data acquisition, comprised two functional runs: (1) a TMS run and (2) a finger tapping task. In addition, an anatomical scan was acquired. A flow diagram of the experiment is shown in Fig. 2a. The TMS protocol (Fig. 2b) comprised a total of twenty stimulation blocks, with five TMS blocks for each of the following intensities: 80%, 90%, 100%, and 110% of the individual's active motor threshold (AMT). TMS blocks were randomized and each stimulation block was followed by a rest condition block of the same duration (10 s). An inhouse written Matlab script controlled TMS timing with respect to the MR scanner trigger signal. The total scanning time for the stimulation paradigm was 6 min 50 s.
263
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Fig. 1. : Measurement setup. (a) Dedicated MR coil array for concurrent TMS/fMRI experiments together with interfacing hardware. (b) TMS/fMRI setup: The MR coil array is placed between TMS and the head of the subject. Photos reproduced with permission from Navarro de Lara et al. (2015).
MR/TMS coil using the standard Siemens console software. This plane was then used to position fMRI slices over the primary motor cortex.
None of the subjects received more than 200 effective TMS pulses per day, and half of them were at subthreshold intensity. The TMS protocol was in full accordance with the present safety guidelines for TMS (Rossi et al., 2009, Siebner et al., 2009).
Transcranial magnetic stimulation and electromyographic recordings Magnetic resonance imaging TMS was performed using a MagPro X100 stimulator and an MRcompatible TMS coil MRi-B91 (Magventure, Farum, Denmark). The stimulator generated biphasic pulses of approximately 280 μs duration. The TMS target region, the representational area of the right first dorsal interosseous (FDI) muscle on the primary motor cortex (M1) was defined for each subject based on the activation maps obtained from the finger tapping experiment during the first MRI session. After analysis of finger tapping data (SPM12), statistical parametric maps and anatomical scans were loaded into a neuronavigation suite (Brainsight 2, Rogue Research Inc., Canada). Subsequently subjects’ heads were registered to their respective anatomical scans using three reference points (nasion, left and right ear). The TMS coil attached to the MR coil array was registered using the manufacturer's calibration block and the coil tracker. Using the neuronavigation tool, positioning of the combined TMS/MR coil array within a distance of less than 5 mm to the target was achieved. Using the EMG integrated in the neuronavigation suite, the motor evoked response of the right FDI muscle was assessed on a grid of 13 points, one at the target (defined by the hotspot of activation maps from the finger tapping paradigm), and at 12 points around the hotspot on 2 rings, 4 points in the first and 8 in the second ring. The ring spacing was 5mm (Parameters for neuronavigation system: 3 rings, circular grid, 7 mm arc length, 5 mm ring spacing). The position with maximum TMS-evoked motor response in the FDI muscle was selected as the target to be used in the TMS/fMRI
Measurements were performed on a 3 T Tim Trio MR Scanner (Siemens Healthcare, Erlangen, Germany). During the first session (using the 32-channel head coil), finger-tapping data was acquired with a standard gradient-echo 2D single-shot EPI sequence with TR / TE = 1800 ms/38 ms, 23 slices, 3 mm slice thickness (1.8 mm slice gap), 1.5×1.5 mm2 in-plane resolution, and MA = 128×128. Anatomical MPRAGE images were also acquired (TR / TE / TI = 2300 ms / 4.21 ms/900 ms, flip angle 9°, 160 slices, 1.25 mm slice thickness, FOV = 240×56 mm2, MA = 240×256, and BW = 238 Hz/pixel). In the second session (using the dedicated MR coil array) functional images were acquired using the CMRR multi-band gradient-echo 2D single-shot EPI sequence (Moeller et al., 2010) with a limited number of slices to allow for higher temporal resolution (TR / TE = 1000 ms/ 33 ms, flip angle 60°, 14 slices, 3 mm slice thickness, 1.5×1.5 mm2 inplane resolution, MA = 128×128, BW = 1502 Hz/pixel and MB factor 2). Slices were aligned parallel to the TMS coil and fully covered the primary motor cortex. The MR coil array (Navarro de Lara, 2015) contained three MR-visible fiducial markers that were used to facilitate positioning the image slab based on a FLASH dataset (TR / TE = 400 ms/2.46 ms, flip angle 41°, 14 slices, 3 mm slice thickness, 0.69×0.69 mm2 in-plane resolution, MA = 256×320, GRAPPA = 2 and BW = 332 Hz/pixel). The three fiducial markers described above were identified in the FLASH images to define a plane parallel to the
Fig. 2. Measurement Protocol. (a) Flow diagram of the TMS experiment. (b) The stimulation protocol consists of 20 blocks, each block starting with 10 seconds baseline (fMRI acquisition but no TMS) followed by 10 seconds TMS (fMRI acquisition and TMS). The stimulation intensity of each block is chosen as one of the four intensities, i.e. 80%, 90%, 100% or 110% of individual's previously determined active motor threshold. (c) fMRI acquisition timing. The yellow blocks corresponds to the acquisition of one slab. fMRI data collection, blue line correspond to TMS stimulation pulse. Volumes are acquired with TR=1 s.
264
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
experiment. It corresponded well to the activations map hotspots. Motor thresholds were defined as the dI/dt values displayed by the stimulator that produced a MEP response between 0.05–0.1 mV within a time frame between 15 and 35 ms after application in the target muscle in two out of four trials during 50% of maximum voluntary contraction. Thresholds were determined outside of the MR environment, shortly before the scanning session. To avoid increase of cortical excitability during this procedure (Pascual-Leone et al., 1994) the stimulation intensity was gradually reduced until no muscle twitch was elicited in the target muscle in half of the trials. The motor threshold was verified in the MR scanner before running the stimulation protocol.
Combined TMS/fMRI The new MR coil is equipped with two sets of markers. The first set consists of three small, liquid-filled capsules that are mounted inside the MR coil's housing. These markers are visible on MR images and define a plane parallel to the MT/TMS coil used to position fMRI acquisition slices. The second set of markers consists of three small reflective spheres mounted on a non-ferromagnetic object and is part of the neuronavigation system. The latter markers are detachable and in this study were removed before moving the subject bed into the scanner. The TMS coil, rigidly combined with the MR coil array, was mounted to a robust coil-holding device (MRi Coil Holder, MagVenture, Farum, Denmark) and positioned using the neuronavigation suite on the redefined target in the scanner room but outside the scanner bore. Coil-orientation was tangential, approximately 45° to the mid-line. Subjects lay supine on an in-house made cushioned head holder, designed for comfortable subject fixation during the experiment. The TMS coil was connected using an extension cable of eight meters through a radiofrequency (RF) filter tube to a filter box mounted on the filter plate. This box also contained hardware to suppress leakage currents to values below 1 µA (Magventure, Farum, Denmark). Due to resistive cable losses and the filter transmission function, the TMS stimulator produces a lower dI/dt for a given stimulation intensity in the scanner, in comparison to the setup where the TMS coil is directly connected to the stimulator outside the scanner. For this reason, the motor thresholds determined outside the scanner had to be recorded as a dI/dt value rather than as percentage of the stimulator output. TMS coil position and the motor thresholds were briefly evaluated before the stimulation experiment by applying single TMS pulses. Head movement was restricted by foampadded cushions and the subject wore ear-plugs to reduce noise throughout the experiment. Acquiring one volume in the TMS/fMRI sequence took 680 ms (Fig. 2c). The minimum TMS delay times to avoid artefacts in the EPI images from TMS pulses were examined in a pilot study where we acquired MR images of a phantom using the same multiband gradientecho 2D single-shot EPI sequence as described above but varied the time after a TMS pulse of 100% stimulation output intensity to the next EPI acquisition. The mean of the first and last 20 image slabs (all without stimulation pulses) was used as a reference image. Twenty-six equidistantly spaced durations between 0 and 100 ms were applied as delay. This procedure was repeated five times. For each delay value, the root mean square (RMS) deviation from the reference image was calculated to obtain a measure of signal dropout. The results of this initial study are shown in Fig. 3c. It can be seen that a delay of 50 ms is sufficient to avoid TMS effects on image quality. In the actual experiment, a much longer delay of 160 ms was chosen, so to place TMS pulses right in the middle of non-acquisition periods (repetition time TR=1000 ms, acquisition time TA=680 ms, time between acquisition TD=320 ms; see also Fig. 2).
Fig. 3. Functional images and determination of minimal TMS delay. Individual functional images acquired with the novel coil during (a) a rest block, and (b) a stimulation block. Slices were acquired parallel to the TMS coil plane. Green arrows indicate the interhemispheric fissure. (c) Percentage of the root mean square error versus time after TMS pulse to the next image acquisition. It can be seen that a minimum delay of 50 ms is sufficient to avoid artifacts or dropouts in the images. For our experiment, we chose a delay time of 160 ms (marked with the blue flash).
Data analysis Data analysis was performed using SPM12b. Pre-processing of fMRI acquisitions included slice-timing correction (Sladky et al., 2011), motion correction and smoothing with a 3 mm FWHM isotropic Gaussian kernel. The pre-processed datasets were analysed using the general lineal model (GLM) as implemented in SPM, i.e. linear regression was performed at each voxel, using generalized least squares with a global approximate AR(1) autocorrelation model. Based on the TMS stimulation intensities (TMS experiment) or movement epochs (finger tapping experiment), models were defined and convolved with SPM's haemodynamic response function. Activation maps for single subjects were obtained by applying a t-test to the GLM parameter estimates obtained for each stimulation intensity or movement epoch. The significance level was set at p < 0.05, family-wise error (FWE) corrected at the voxel level. For visualisation purposes, anatomical images acquired with the 32-channel head coil were co-registered to the anatomical images acquired with the 7-channel MR coil array with an in-house written software in MATLAB, which applied N4 bias correction (Tustison et al., 2010) prior to performing rigid registration using SPM. In order to calculate the coordinates of the point where the central axis of the TMS coil penetrated the cortex, a skull-stripped version of the co-registered anatomical data (acquired with the 32-channel head coil) using the AFNI tool 3dSkullStrip was generated. Then, the intersection of brain surface and TMS coil axis as obtained from three 265
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Single subject results
markers placed in the co-planar MR coil was calculated with Matlab code written in-house. These subject-space coordinates were transformed to MNI space by using the unified segmentation routine as implemented in SPM12 applied to the co-registered 32-channel anatomical data. For each subject, three voxels were defined, all based on the activation maps from the 110% stimulation amplitude blocks: (i) global maximum voxel (activation maximum over the whole brain); (ii) Finger-Tapping-M1 maximum voxel (activation maximum within the M1 hand area as obtained from the individual's finger tapping activation map); (iii) MNI-M1 maximum voxel (activation maximum within the M1 hand area as defined by transforming the hand knob MNI coordinate [−40 −20 52] (Cardenas-Morales et al., 2014) to subject space). For (ii), a search radius of 2 cm was applied to determine the voxel of maximum M1 activation. For (iii), the “find local maximum” functionality as implemented in SPM12 was used, starting from MNI coordinate [−40 −20 52]. BOLD signal intensity changes for each subject were calculated from the parameter estimates obtained by SPM divided by the mean signal (constant term in the design matrix). Mean signal time courses for finger tapping and all stimulation intensities were calculated as a percentage with respect to the last two time points of each experimental block (Bestmann et al., 2004).
Activation maps of a typical subject are shown in Fig. 4. Maps show activation from the finger tapping experiment (Fig. 4a) and during TMS with 110% (Fig. 4b, d) and 100% aMT stimulation levels (Fig. 4c, e). Each map is centred on the respective maximum, i.e. maximum finger tapping activation (Fig. 4a), global maximum (Fig. 4b, c), and M1 maximum (Fig. 4d, e). All maps are thresholded at p < 0.05 (voxel-level FWE-corrected). Table 1 shows the activation t-values and the parameter estimates at the three extracted voxels for all subjects and for all stimulation intensities. The FWE corrected and uncorrected p-values are also listed in the table. For the subject shown in Fig. 4, global activation maxima were statistically significant (p < 0.05, FWE-corrected) for all TMS intensities, while M1 maxima were found significant only for 100% and 110% motor threshold stimulation intensity (see Table 1, Subject 4). For all subjects, global maximal activations were found below the coil center.
Group results To evaluate coil sensitivity, we focused on the quality of the acquired signal in terms of BOLD signal intensity change and statistical significance. For this purpose, group-level results were calculated as the mean of the percentage BOLD signal intensity change for the three voxels defined above. The results are plotted in Fig. 5 for each stimulation intensity as mean and standard error of the mean (SEM) values. Across the group, signal changes (mean ± SEM, [%]) at the global maximum were 0.88 ± 0.24, 1.09 ± 0.29, 1.65 ± 0.43, 2.77 ± 0.73 for 80%, 90%, 100%, and 110%, respectively. Corresponding values in local M1 maxima based on the finger tapping results were 0.43 ± 0.30, 0.63 ± 0.30, 1.01 ± 0.23, and 2.01 ± 0.43, respectively. For MNI-based maxima signal changes were calculated as 0.73 ± 0.28, 0.91 ± 0.37, 1.34 ± 0.49 and 2.21 ± 0.61, respectively. Across the group, signal changes in the global maximum were statistically significant for all stimulation intensities. At the voxel of M1 maxima only values for the 100% and 110% stimulation intensities were statistically significant across subjects. In both graphs, an ascending monotonic relationship of mean signal changes over stimulation intensity can be seen.
Results None of the subjects reported any adverse side effects after participation. Functional images acquired with the new dedicated MR coil array are shown in Fig. 3a and b in both states, during stimulation and rest blocks. None of the volumes acquired during the stimulation protocol were affected by TMS pulses or presented dropouts under the coil. From the measurements in the phantom, we concluded that ensuring a wait time of at least 50 ms before EPI acquisition after a TMS pulse, no distortions or artefacts on the functional images were produced either by the magnetic field of the TMS pulses or possible vibrations (see Fig. 3c). No susceptibility effects due to the presence of the TMS system were observed.
Fig. 4. Single subject results (subject 4). Arrows point to the activation peak used for extracting parameters estimates and time courses. (a) Activation map obtained from the finger tapping experiment. (b)-(e) TMS-induced activation for 110% and 100% aMT stimulation at the global maximum (b-c) and the local maximum next to the M1 area as defined via the finger tapping results (d-e).
266
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Table 1 Peak activation t-values, parameter estimates, voxel-wise FWE-corrected and uncorrected p-values at the three extracted voxels, i.e., the global activation maximum, the local activation maximum near M1 based on the individual finger tapping localizer, and the local activation maximum near M1 based on fixed MNI coordinates, respectively. Global Activation Maximum
Subject 1
Subject 2
Subject 3
Subject 4
Subject 5
Subject 6
Subject 7
Levels
t-Value
beta
p corr. (FWE)
p unc.
Local Activation Maximum near M1 based on Finger Tapping t-Value beta p corr. (FWE) p unc.
110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80
15.36 9.11 3.97 7.43 8.92 6.67 5.14 3.71 10.38 8.83 10.26 8.87 19.24 15.58 10.83 8.18 19.87 10.97 2.52 0.49 11.59 6.24 5.92 4.82 15.32 3.69 3.57 1.45
11.2 6.51 3.16 5.02 11.15 7.94 5.75 5.72 1.84 1.55 1.67 1.61 21.19 16.1 12.17 8.92 17.23 8.97 2.26 0.42 6.85 3.6 3.39 2.96 7.51 1.88 1.92 0.79
< 10-10 < 10-10 1 < 10-7 < 10-10 < 10-5 0.048 1 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-9 < 10-10 < 10-10 1 1 < 10-10 < 10-3 < 10-3 0.24 < 10-10 1 1 1
< 10-16 < 10-16 < 10-4 < 10-12 < 10-16 < 10-10 < 10-6 < 10-3 < 10-16 < 10-14 < 10-16 < 10-16 < 10-16 < 10-16 < 10-16 < 10-4 < 10-16 < 10-16 0.0061 0.31 < 10-16 < 10-9 < 10-8 < 10-5 < 10-16 < 10-3 < 10-3 0.074
15.36 9.11 3.97 7.43 7.32 5.02 3.33 2.04 10.38 8.83 10.26 8.87 16 8.25 5.11 -0.21 15.54 5.85 -2.66 -4.15 11.59 6.24 5.92 4.82 12.04 1.2 0.67 0.37
< 10-10 < 10-10 1 < 10-7 < 10-6 0.087 1 1 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-9 0.057 1 < 10-10 0.0012 1 1 < 10-10 < 10-3 < 10-3 0.24 < 10-10 1 1 1
11.2 6.51 3.16 5.02 5.93 3.87 2.41 2.04 1.84 1.55 1.67 1.61 12.06 5.83 3.93 -0.15 10.65 3.78 -1.88 -2.85 6.85 3.6 3.39 2.96 5 0.51 0.31 0.16
< 10-16 < 10-16 < 10-4 < 10-12 < 10-12 < 10-6 < 10-3 0.021 < 10-16 < 10-14 < 10-16 < 10-16 < 10-16 < 10-8 < 10-4 0.58 < 10-16 < 10-8 1 1 < 10-16 < 10-9 < 10-8 < 10-5 < 10-16 0.12 0.25 0.36
Local Activation Maximum near M1 based on MNI t-Value beta p corr. p unc. (FWE) 10.08 2.89 0.85 2.28 8.92 6.67 5.14 3.71 6.88 2.46 5.78 4.99 16 8.25 5.11 -0.21 6.85 2.32 -0.99 -0.48 6.26 4.32 1.40 4.15 6.76 0.04 -0.60 -1.08
6.25 1.76 0.58 1.31 11.15 7.94 5.75 5.72 3.09 1.1 2.38 2.29 12.06 5.83 3.93 -0.15 7.27 2.32 -1.08 -0.51 3.11 2.09 0.68 2.14 2.46 0.02 -0.24 -0.44
< 10-10 1 1 1 < 10-10 < 10-5 0.048 1 < 10-5 1 0.0017 0.1 < 10-10 < 10-9 0.057 1 < 10-5 1 1 1 < 10-3 0.98 1 1 < 10-5 1 1 1
< 10-16 0.002 0.2 0.01 < 10-16 < 10-10 < 10-6 < 10-3 < 10-10 0.0072 < 10-8 < 10-6 < 10-16 < 10-8 < 10-4 0.58 < 10-10 0.01 0.84 0.68 < 10-9 < 10-5 0.081 < 10-4 < 10-10 0.48 0.73 0.86
Note that the FWE-correction implementation used limited p-value accuracy to < 10–10.
Fig. 5. Group results for M1 after TMS stimulation. Mean BOLD signal change and SEM over the whole group (n=7) across the different stimulation intensities in the voxel of (a) global maximal activation, (b) local maximum activation near M1 based on the FT results, and c) local maximum activation near M1 based on the MNI coordinates. Results marked with an asterisk are statistically significant (p < 0.05). Error bars indicate standard errors of the means. It can be seen that activation increases with stimulation amplitude in all regions examined. At the voxel of global maximum activation, activation was statistically significant for all stimulation intensities. Local maxima were significant for 100% and 110% aMT stimulation only.
caused signal increase of 1.3% (1.0%), 0.8% (0.5%) and 0.6% (0.3%) for global, finger tapping and MNI maxima, respectively. The presented values are slightly different than the values calculated using the parameter estimates as obtained with SPM. This difference can be explained by the different approaches used for the baseline signal calculation. Mean MNI coordinates and spread for the reported peaks were calculated together with the projection of the TMS coil centre onto the brain. The spread was defined as the mean of the standard deviation in
In order to examine TMS-induced activation changes in more detail, time courses were extracted and averaged over repeated blocks of the same intensity and then averaged across subjects. Fig. 6 shows the resulting plots for the global maximum (a) and the local maxima based on finger tapping (b) and MNI-coordinate (c). Time courses for the finger tapping runs are displayed for comparison. For 110% stimulation intensity we obtain 3.3%, 2.4%, 1.5% for global, finger tapping and MNI maxima, respectively. 100% stimulation resulted in 2.1%, 1.4% and 0.9%. Subthreshold stimulation with 90% (80%)
267
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
primary motor cortex using four different stimulation intensities. Our results show that observed BOLD responses to TMS are considerably higher compared to previous publications where the standard birdcage coil setup was used (Baudewig et al., 2001; Bestmann et al., 2004). It seems likely that this is due to the greatly improved sensitivity due to the new setup. A monotonically ascending relationship of BOLD signal change with TMS intensity has been obtained in all three regions examined, i.e. global maximum voxel, Finger-Tapping-M1 maximum voxel, MNI-M1 maximum voxel. Only the global maximum voxel showed statistically significant signal changes for all stimulation intensities. Consistent with results obtained in other studies (Bohning et al., 1999, Baudewig et al., 2001, Bestmann et al., 2003b, 2004), the results on the local maxima next to M1 were found to be significant for the 100% and 110% stimulation intensities. At this particular M1 site, earlier physiological studies (Fisher et al., 2002) proposed that intensities under motor threshold affect low-threshold inhibitory nodes, which produce inhibitory synaptic activity within the primary motor cortex (Bestmann et al., 2003b). A sigmoidal effect between TMS intensities and activation directly at M1 has been proposed based on these effects (Bestmann et al., 2003b, Hanakawa et al., 2009). It is important to note that no susceptibility effects or distortions of the images were observed beneath the coil in the fMRI data. We suggest that BOLD changes detected below the coil, i.e., the stimulation site, reflect an intensity-dependent increase in local neuronal activity even below the individual motor threshold. On the other hand, haemodynamic response at M1 was observed at threshold levels and above, while responses to subthreshold stimulation were below significance level, possibly due to compensatory effects caused by subcomponents of the motor network. Fig. 6 shows that the acquired signal returns to baseline within the fixed 10 s rest period of our stimulation protocol. Carry-over effects may still have occurred and could explain the variance seen for low intensities. In this study, the 2 out of 4 rule was used to determine the motor threshold. This approach differs from the established 5 out of 10 rule, and might have caused a slight misestimation of threshold levels. The results obtained in this study also demonstrate the applicability of the novel setup to acquire high-quality functional images during stimulation. The time courses presented are now comparable to BOLD signal changes obtained in typical fMRI block-design experiments, with peaks of about 3% signal change. A limitation of previous TMS/fMRI setups concerned TMS coil positioning. In the new setup, positioning the TMS coil is greatly facilitated, as mounting the TMS coil is no longer confined to the limited space within a birdcage coil, enhancing flexibility. Also, we have shown here that neuronavigation tools can now be directly applied, making this cumbersome task easier and more accurate. It also opens up the possibility to perform concurrent TMS/fMRI studies in new target regions with no immediate sensory output. In addition to mapping local TMS-induced activation, one of the main goals of neuroimaging studies combined with brain stimulations techniques is to investigate the distributed effects it can have on functional brain networks (Bestmann and Feredoes, 2013). For this reason the assessment of remote effects of the stimulation represents an important aspect that has to be fulfilled by the neuroimaging technique. The proposed set up with one 7-channel surface coil array is limited in this way. However, in order to answer these more complex research questions, such as TMS-remote effects, we propose the use of an additional MR coil array positioned opposite to the site of stimulation. Using the combination of the two MR coil arrays with advanced fast and precise imaging and neuronavigation methods, our group has already demonstrated acute local and network effects of TMS over the left DLPFC (Tik et al., 2016). Using the newly developed hardware comprising seven RF receive coil elements, tightly fixed to the TMS coil, it is now possible to benefit from parallel and multiband imaging approaches, thereby improving
Fig. 6. : Group-averaged BOLD signal time courses averaged over blocks Stimulation blocks are indicated in green, and time courses extracted from finger tapping data are shown as dashed lines. Colors correspond to stimulation amplitudes: 80% (lilac), 90% (yellow), 100% (orange), 110% (blue). Data is plotted for the global maximum (a) and the local M1 maxima based on finger tapping (b) and MNI-coordinate (c). Table 2 Mean MNI coordinates and mean of the standard deviation given as the spread. These values were calculated for the reported peaks and for the voxel found as the projection of the TMS coil center on the brain. Voxel Specification
MNI Coordinates [x, y, z] Mean
Standard Deviation
TMS coil center projection on brain surface
[−40, −15, 64]
[9.3, 11.9, 7.4]
Voxel at global activation maximum
[−40, −17, 59] [−42, −24, 55] [−38, −15, 54]
[6.2, 12.4, 8.1]
Voxel at local activation maximum near M1 (based on finger tapping) Voxel at local activation maximum near M1 (MNI based)
[4.3, 10.7, 4.1] [3.1, 4.5, 3.6]
each direction. The respective coordinates are summarised in Table 2. Discussion Our results clearly demonstrate that the new setup using a novel, dedicated MR coil array can be successfully applied to conduct concurrent high sensitivity TMS/fMRI experiments with high temporal and spatial resolution. The 7-channel MR receive array has already been shown to boost sensitivity compared to the standard birdcage coil approach (Navarro de Lara et al., 2015). In this study we show how this benefit in sensitivity can be used to assess TMS-induced activity in the 268
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Denslow, S., Lomarev, M., George, M.S., Bohning, D.E., 2005. Cortical and subcortical brain effects of transcranial magnetic stimulation (TMS)-induced movement: an interleaved TMS/functional magnetic resonance imaging study. Biol. Psychiatry 57, 752–760. Fisher, R.J., Nakamura, Y., Bestmann, S., Rothwell, J.C., Bostock, H., 2002. Two phases of intracortical inhibition revealed by transcranial magnetic threshold tracking. Exp. Brain Res. 143, 240–248. Hanakawa, T., Mima, T., Matsumoto, R., Abe, M., Inouchi, M., Urayama, S., Anami, K., Honda, M., Fukuyama, H., 2009. Stimulus-response profile during single-pulse transcranial magnetic stimulation to the primary motor cortex. Cereb. Cortex 19, 2605–2615. Lefaucheur, J.P., Andre-Obadia, N., Antal, A., Ayache, S.S., Baeken, C., Benninger, D.H., Cantello, R.M., Cincotta, M., de Carvalho, M., De Ridder, D., Devanne, H., Di Lazzaro, V., Filipovic, S.R., Hummel, F.C., Jaaskelainen, S.K., Kimiskidis, V.K., Koch, G., Langguth, B., Nyffeler, T., Oliviero, A., Padberg, F., Poulet, E., Rossi, S., Rossini, P.M., Rothwell, J.C., Schonfeldt-Lecuona, C., Siebner, H.R., Slotema, C.W., Stagg, C.J., Valls-Sole, J., Ziemann, U., Paulus, W., Garcia-Larrea, L., 2014. Evidence-based guidelines on the therapeutic use of repetitive transcranial magnetic stimulation (rTMS). Clin. Neurophysiol. 125, 2150–2206. Moeller, S., Yacoub, E., Olman, C.A., Auerbach, E., Strupp, J., Harel, N., Ugurbil, K., 2010. Multiband multislice GE-EPI at 7 T, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magn. Reson. Med. 63, 1144–1153. Moisa, M., Siebner, H.R., Pohmann, R., Thielscher, A., 2012. Uncovering a contextspecific connectional fingerprint of human dorsal premotor cortex. J. Neurosci.: Off. J. Soc. Neurosci. 32, 7244–7252. Navarro de Lara, L.I., Windischberger, C., Kuehne, A., Woletz, M., Sieg, J., Bestmann, S., Weiskopf, N., Strasser, B., Moser, E., Laistler, E., 2015. A novel coil array for combined TMS/fMRI experiments at 3 T. Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med. / Soc. Magn. Reson. Med. 74, 1492–1501. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97–113. Pascual-Leone, A., Valls-Sole, J., Wassermann, E.M., Hallett, M., 1994. Responses to rapid-rate transcranial magnetic stimulation of the human motor cortex. Brain: J. Neurol. 117 (Pt 4), 847–858. Rossi, S., Hallett, M., Rossini, P.M., Pascual-Leone, A., Grp, S.T.C., 2009. Safety, ethical considerations, and application guidelines for the use of transcranial magnetic stimulation in clinical practice and research. Clin. Neurophysiol. 120, 2008–2039. Rossini, P., Rossi, S., 2007. Transcranial magnetic stimulation: diagnostic, therapeutic, and research potential. Neurology 68, 484–488. Ruff, C.C., Bestmann, S., Blankenburg, F., Bjoertomt, O., Josephs, O., Weiskopf, N., Deichmann, R., Driver, J., 2008. Distinct causal influences of parietal versus frontal areas on human visual cortex: evidence from concurrent TMS-fMRI. Cereb. Cortex 18, 817–827. Ruff, C.C., Blankenburg, F., Bjoertomt, O., Bestmann, S., Freeman, E., Haynes, J.D., Rees, G., Josephs, O., Deichmann, R., Driver, J., 2006. Concurrent TMS-fMRI and psychophysics reveal frontal influences on human retinotopic visual cortex. Curr. Biol.: CB 16, 1479–1488. Ruff, C.C., Blankenburg, F., Bjoertomt, O., Bestmann, S., Weiskopf, N., Driver, J., 2009. Hemispheric differences in frontal and parietal influences on human occipital cortex: direct confirmation with concurrent TMS-fMRI. J. Cogn. Neurosci. 21, 1146–1161. Shitara, H., Shinozaki, T., Takagishi, K., Honda, M., Hanakawa, T., 2011. Time course and spatial distribution of fMRI signal changes during single-pulse transcranial magnetic stimulation to the primary motor cortex. NeuroImage 56, 1469–1479. Siebner, H.R., Bergmann, T.O., Bestmann, S., Massimini, M., Johansen-Berg, H., Mochizuki, H., Bohning, D.E., Boorman, E.D., Groppa, S., Miniussi, C., PascualLeone, A., Huber, R., Taylor, P.C., Ilmoniemi, R.J., De Gennaro, L., Strafella, A.P., Kahkonen, S., Kloppel, S., Frisoni, G.B., George, M.S., Hallett, M., Brandt, S.A., Rushworth, M.F., Ziemann, U., Rothwell, J.C., Ward, N., Cohen, L.G., Baudewig, J., Paus, T., Ugawa, Y., Rossini, P.M., 2009. Consensus paper: combining transcranial stimulation with neuroimaging. Brain Stimul. 2, 58–80. Silvanto, J., Pascual-Leone, A., 2012. Why the assessment of causality in brain-behavior relations requires brain stimulation. J. Cogn. Neurosci. 24, 775–777. Sladky, R., Friston, K.J., Trostl, J., Cunnington, R., Moser, E., Windischberger, C., 2011. Slice-timing effects and their correction in functional MRI. Neuroimage 58, 588–594. Tik M., Woletz M., Navarro de Lara L., Sladky R., Hoffmann A., Hummer A., Windischberger C., 2016. Multimodal Assessment of Local and Remote TMSinduced Effects. In: Annual Meeting of the Organization for Human Brain Mapping Geneva, Switzerland. Tustison, N.J., Avants, B.B., Cook, P.A., Zheng, Y., Egan, A., Yushkevich, P.A., Gee, J.C., 2010. N4ITK: improved N3 bias correction. IEEE Trans. Med. Imag. 29, 1310–1320. Walsh, V., Pascual-Leone, A., 2005. Transcranial Magnetic Stimulation: a Neurochronometrics of Mind. MIT Press, Cambridge, MA. Weiskopf, N., Josephs, O., Ruff, C.C., Blankenburg, F., Featherstone, E., Thomas, A., Bestmann, S., Driver, J., Deichmann, R., 2009. Image artifacts in concurrent transcranial magnetic stimulation (TMS) and fMRI caused by leakage currents: modeling and compensation. J. Magn. Reson. Imag.: JMRI 29, 1211–1217. Wiggins, G.C., Triantafyllou, C., Potthast, A., Reykowski, A., Nittka, M., Wald, L.L., 2006. 32-channel 3 T receive-only phased-array head coil with soccer-ball element geometry. Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med. / Soc. Magn. Reson. Med. 56, 216–223.
temporal and spatial resolution. Until now, these new acquisition techniques, which have become standard in current fMRI experiments, were not available to concurrent TMS/fMRI studies. The improvement in sensitivity, which in turn can yield higher spatial resolution, together with better temporal resolution by applying parallel imaging techniques may help to improve the specificity of concurrent TMS/fMRI experiments. Taken together, here we show the applicability of neuronavigated concurrent TMS/fMRI using a new dedicated high-sensitivity coil array for precise mapping of local intensity-dependent BOLD responses even at an individual subject's level. This flexible setup will allow for future experiments targeting different cortical targets to understand TMS mechanism of action and improve therapeutic effects of TMS in clinical settings. Acknowledgements We thank Gunnar Hallsson, Claus Mathiesen and colleagues from MagVenture (Denmark) for their support in performing this study. This work has been financially supported by the HRSM Austrian BMWFJ FFG Project Nr. 832107, “Vienna Research Studio for Ultra-High Field Magnetic Resonance Applications”. References Auerbach, E.J., Xu, J., Yacoub, E., Moeller, S., Ugurbil, K., 2013. Multiband accelerated spin-echo echo planar imaging with reduced peak RF power using time-shifted RF pulses. Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med. / Soc. Magn. Reson. Med. 69, 1261–1267. Baudewig, J., Siebner, H., Bestmann, S., Tergau, F., Tings, T., Paulus, W., Frahm, J., 2001. Functional MRI of cortical activations induced by transcranial magnetic stimulation (TMS). Neuroreport 12, 3543–3548. Bestmann, S., Baudewig, J., Frahm, J., 2003a. On the synchronization of transcranial magnetic stimulation and functional echo-planar imaging. J. Magn. Reson. imaging: JMRI 17, 309–316. Bestmann, S., Baudewig, J., Siebner, H.R., Rothwell, J.C., Frahm, J., 2003b. Subthreshold high-frequency TMS of human primary motor cortex modulates interconnected frontal motor areas as detected by interleaved fMRI-TMS. NeuroImage 20, 1685–1696. Bestmann, S., Baudewig, J., Siebner, H.R., Rothwell, J.C., Frahm, J., 2004. Functional MRI of the immediate impact of transcranial magnetic stimulation on cortical and subcortical motor circuits. Eur. J. Neurosci. 19, 1950–1962. Bestmann S., Feredoes E., 2013. Combined neurostimulation and neuroimaging in cognitive neuroscience: past, present, and future. Annals of the New York Academy of Sciences 1296, pp. 11–30. 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 with concurrent TMS-fMRI. Neuropsychologia 44, 2959–2971. Bestmann, S., Swayne, O., Blankenburg, F., Ruff, C.C., Haggard, P., Weiskopf, N., Josephs, O., Driver, J., Rothwell, J.C., Ward, N.S., 2008. Dorsal premotor cortex exerts state-dependent causal influences on activity in contralateral primary motor and dorsal premotor cortex. Cereb. Cortex 18, 1281–1291. Bohning, D., Shastri, A., McConell, K., Nahas, Z., Lorberbaum, J., Roberts, D., Teneback, C., Vincent, D., George, M., 1999. A combined TMS/fMRI Study of intensitydependent TMS over motor cortex. Biol. Psychiatry 45, 385–394. Bohning, D., Shastri, A., Nahas, Z., JP, L., Andersen, S., Dannels, W., Haxthausen, E., Vincent, D., George, M., 1998. Echoplanar BOLD fMRI of brain activation induced by concurrent transcranial magnetic stimulation. Investig. Radiol. 33, 336–340. Bohning, D.E., Shastri, A., Lomarev, M.P., Lorberbaum, J.P., Nahas, Z., George, M.S., 2003. BOLD-fMRI response vs. transcranial magnetic stimulation (TMS) pulse-train length: testing for linearity. J. Magn. Reson Imag. 17, 279–290. Bohning, D.E., Shastri, A., Wassermann, E.M., Ziemann, U., Lorberbaum, J.P., Nahas, Z., Lomarev, M.P., George, M.S., 2000. BOLD-f MRI response to single-pulse transcranial magnetic stimulation (TMS). J. Magn. Reson. Imag.: JMRI 11, 569–574. Bungert, A., Chambers, C.D., Phillips, M., Evans, C.J., 2012. Reducing image artefacts in concurrent TMS/fMRI by passive shimming. NeuroImage 59, 2167–2174. Cardenas-Morales, L., Volz, L.J., Michely, J., Rehme, A.K., Pool, E.M., Nettekoven, C., Eickhoff, S.B., Fink, G.R., Grefkes, C., 2014. Network connectivity and individual responses to brain stimulation in the human motor system. Cereb. Cortex 24, 1697–1707. Chen, R., Cros, D., Curra, A., Di Lazzaro, V., Lefaucheur, J., Magistris MaZ, U., 2008. The clinical diagnostic utility of transcranial magnetic stimulation: report of an IFCN committee. Clin. Neurophysiol.: Off. J. Int. Fed. Clin. Neurophysiol. 119 (3), 504–532.
269
Contents lists available at ScienceDirect
NeuroImage journal homepage: www.elsevier.com/locate/neuroimage
High-sensitivity TMS/fMRI of the Human Motor Cortex Using a Dedicated Multichannel MR Coil
MARK
Lucia I. Navarro de Laraa,b,1, Martin Tika,b,1, Michael Woletza,b, Roberta Frass-Kriegla,b, ⁎ Ewald Mosera,b, Elmar Laistlera,b, Christian Windischbergera,b, a b
Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Währinger Guertel 18–20, A-1090 Wien, Vienna, Austria MR Center of Excellence, Medical University of Vienna, Vienna, Austria
A R T I C L E I N F O
A BS T RAC T
Keywords: TMS fMRI Coil array parallel imaging motor cortex concurrent TMS/fMRI
Purpose: To validate a novel setup for concurrent TMS/fMRI in the human motor cortex based on a dedicated, ultra-thin, multichannel receive MR coil positioned between scalp and TMS system providing greatly enhanced sensitivity compared to the standard birdcage coil setting. Methods: A combined TMS/fMRI design was applied over the primary motor cortex based on 1 Hz stimulation with stimulation levels of 80%, 90%, 100%, and 110% of the individual active motor threshold, respectively. Due to the use of a multichannel receive coil we were able to use multiband-accelerated (MB=2) EPI sequences for the acquisition of functional images. Data were analysed with SPM12 and BOLD-weighted signal intensity time courses were extracted in each subject from two local maxima (individual functional finger tapping localiser, fixed MNI coordinate of the hand knob) next to the hand area of the primary motor cortex (M1) and from the global maximum. Results: We report excellent image quality without noticeable signal dropouts or image distortions. Parameter estimates in the three peak voxels showed monotonically ascending activation levels over increasing stimulation intensities. Across all subjects, mean BOLD signal changes for 80%, 90%, 100%, 110% of the individual active motor threshold were 0.43%, 0.63%, 1.01%, 2.01% next to the individual functional finger tapping maximum, 0.73%, 0.91%, 1.34%, 2.21% next to the MNI-defined hand knob and 0.88%, 1.09%, 1.65%, 2.77% for the global maximum, respectively. Conclusion: Our results show that the new setup for concurrent TMS/fMRI experiments using a dedicated MR coil array allows for high-sensitivity fMRI particularly at the site of stimulation. Contrary to the standard birdcage approach, the results also demonstrate that the new coil can be successfully used for multibandaccelerated EPI acquisition. The gain in flexibility due to the new coil can be easily combined with neuronavigation within the MR scanner to allow for accurate targeting in TMS/fMRI experiments.
Introduction Transcranial magnetic stimulation (TMS) is a non-invasive brain stimulation method widely applied in scientific research and a promising clinical diagnostic tool (Walsh and Pascual-Leone, 2005, Rossini and Rossi, 2007, Chen et al., 2008) and treatment option (Lefaucheur et al., 2014). It allows for interacting with the function of brain regions at well-defined time points and thus, establishment of causal relationships between the region stimulated and behaviour (Silvanto and Pascual-Leone, 2012). The combination of TMS with functional magnetic resonance imaging (fMRI) has received considerable attention over the last years because it promises to provide a better
⁎
1
understanding of the mechanisms underlying TMS effects. So-called “online” approaches, where TMS stimulation is applied to subjects during fMRI acquisition, are of particular interest. Although early pioneering work already demonstrated the feasibility of this combination (Bohning et al., 1998, Bohning et al., 1999), online TMS/fMRI still poses a number of challenges. The current state-of-the-art setup for concurrent TMS/fMRI uses a large-volume head coil, usually a large birdcage coil, where the MRcompatible TMS coil is mounted inside the volume coil. This setup has been used to explore mainly the primary motor cortex as M1 stimulation results can be observed as visible finger twitches and quantified as motor evoked potentials (MEPs) using Electromyography (EMG).
Corresponding author at: Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Währinger Guertel 18–20, A-1090 Wien, Vienna, Austria. E-mail address: [email protected] (C. Windischberger). Contributed equally.
http://dx.doi.org/10.1016/j.neuroimage.2017.02.062 Received 23 June 2016; Accepted 21 February 2017 Available online 22 February 2017 1053-8119/ © 2017 Elsevier Inc. All rights reserved.
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
coils) were accurately studied demonstrating the absence of significant induction currents in the elements of the imaging coil array during TMS stimulation. Another point addressed in the publication was the difference in quality and quantity of the field intensity produced by the TMS with and without the presence of the MR coil between the TMS and a phantom. It was shown that the only effect of the MR coil presence was a loss of not more than 15% of the field intensity, without producing any distortion in TMS field distribution. In addition, the multi-channel receiver approach with the novel MR coil allowed acquisition at high temporal and spatial resolution employing parallel imaging and multi-band techniques. In the present study we aimed for demonstrating the benefits of the novel RF coil setup in concurrent TMS/fMRI experiments focused on the primary motor cortex. We acquired data with unprecedented high spatial and temporal resolution. The absence of a birdcage coil also allowed for straight-forward application of neuronavigation tools inside the scanner as no surrounding volume coil limited the visibility of retro-reflective markers. From a data acquisition perspective, we tested the use of multi-band (also referred to as simultaneous multi-slice, SMS) imaging with the new coil setup to benefit from significant scan time reduction (Auerbach et al., 2013). Based on the expected gain in sensitivity we also aimed for analysing the relationship between the TMS-induced BOLD signal change amplitudes and the TMS intensities applied at the stimulation site.
Several studies have been conducted using this setup focusing on local effects and motor network modulations. For other experiments where the observed region was not located directly under the TMS coil, a surface coil placed above the investigated region was used to increase fMRI sensitivity. This setup was successfully used for visual cortex experiments where stimulation was applied to parietal and frontal regions (Ruff et al., 2006, Ruff et al., 2008, Ruff et al., 2009). The first experiments applied TMS at a frequency of 1 Hz over the left motor cortex, with either changing stimulation intensities (Bohning et al., 1999) or pulse train lengths (Bohning et al., 2003). Extension to full brain coverage resulted in activation of subcortical areas for this particular TMS frequency (Denslow et al., 2005). It was also shown that higher stimulation frequencies are feasible via inter-slice stimulation approaches (Baudewig et al., 2001, Bestmann et al., 2003b, 2004). When applying subthreshold stimulation, significant signal changes in remote regions were detected in anatomically interconnected areas to stimulation target primary motor cortex (M1), such as supplementary motor area (SMA), cingulate motor region, and left dorsal premotor cortex (Baudewig et al., 2001, Bestmann et al., 2003b, 2004). Surprisingly, no local activation changes were observed at stimulation below motor threshold. Since pulses above but not below motor threshold led to measurable changes in BOLD (Blood oxygenation level dependent) directly at M1, it was proposed that activation in this region for supra-threshold pulses could be influenced by reafferent feedback from the muscle twitch (Baudewig et al., 2001). BOLD response to single TMS pulses over the motor cortex has been investigated using interleaved TMS/fMRI (Bohning et al., 2000, Bestmann et al., 2006, Hanakawa et al., 2009, Shitara et al., 2011). They validated the use of the hemodynamic response function for TMSinduced fMRI changes and proposed linear and non-linear models for different brain regions as response to different TMS stimulation intensities. Other studies investigated the functional interaction between dorsal premotor cortex and interconnected motor regions during different motor tasks (Bestmann et al., 2008, Moisa et al., 2012). In these experiments, TMS was applied for perturbing an on-going task in order to measure local and remote effects. For a general review of combining TMS with neuroimaging see e.g. (Siebner et al., 2009, Bestmann and Feredoes, 2013). Important efforts were made to resolve various technical problems presented by the combination of TMS and fMRI. Artefacts in echoplanar imaging can be minimized if TMS pulses are applied at least 100 ms before EPI acquisition and no TMS pulses are applied during imaging (Bestmann et al., 2003a). Leakage-current artefacts produced by the TMS coil can be minimized using a relay-diode combination (Weiskopf et al., 2009). Susceptibility effects due to the TMS coil were compensated using passive shimming in order to avoid image distortions and signal dropouts (Bungert et al., 2012). However, the most crucial challenge in these experiments is caused by the limitation in signal-to-noise ratio (SNR) due to the use of largediameter birdcage coils, resulting in poor MR sensitivity. A birdcage coil is usually limited to a single (quadrature) RF channel and achieves considerably lower signal-to-noise ratio compared to the current stateof-the-art multi-channel receive coil arrays (Wiggins et al., 2006). In addition, TMS application is severely hampered by the space limits within birdcage coils, reducing patient comfort and positioning choices. To overcome these limitations we developed a dedicated MR multi-channel receive coil array for concurrent TMS/fMRI experiments (Navarro de Lara et al., 2015) shown in Fig. 1a, which allows for very flexible TMS positioning (see Fig. 1b). There it was shown that the new setup yields more than 5-times higher SNR at a target depth of 3 cm compared to the state-of-the-art birdcage coil approach. Interactions between both systems (i.e. TMS and RF
Methods Subjects Seven young healthy subjects (all females, age: 26.1 ± 3.6 years) participated in the experiments after giving written informed consent. None of them reported any history of neuropsychiatric disease and no clinical evidence of neurological dysfunction. All subjects were consistently right-handed as assessed with the Edinburgh Handedness Inventory EHI (Oldfield, 1971). The mean handedness score on the EHI was 83 (SD 13). The study was approved by the ethics board of the Medical University of Vienna and conducted according to the Declaration of Helsinki. Experimental protocol The experiment consisted of two sessions. The manufacturer's standard 32-channel head coil (Siemens, Erlangen, Germany) was used in the first session, while in the second session data were acquired using the novel dedicated MR coil array for combined TMS/fMRI, developed by our group (Navarro de Lara et al., 2015). In the first session where data was acquired with the standard 32channel head coil, subjects were asked to perform a standard finger tapping paradigm using the right hand (nine 10 s blocks each separated by 10 s rest) and an anatomical scan was acquired for reference. The results of this session were used as neuronavigation-based target in the following TMS/fMRI session. The second session, where the new receive-array was used for data acquisition, comprised two functional runs: (1) a TMS run and (2) a finger tapping task. In addition, an anatomical scan was acquired. A flow diagram of the experiment is shown in Fig. 2a. The TMS protocol (Fig. 2b) comprised a total of twenty stimulation blocks, with five TMS blocks for each of the following intensities: 80%, 90%, 100%, and 110% of the individual's active motor threshold (AMT). TMS blocks were randomized and each stimulation block was followed by a rest condition block of the same duration (10 s). An inhouse written Matlab script controlled TMS timing with respect to the MR scanner trigger signal. The total scanning time for the stimulation paradigm was 6 min 50 s.
263
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Fig. 1. : Measurement setup. (a) Dedicated MR coil array for concurrent TMS/fMRI experiments together with interfacing hardware. (b) TMS/fMRI setup: The MR coil array is placed between TMS and the head of the subject. Photos reproduced with permission from Navarro de Lara et al. (2015).
MR/TMS coil using the standard Siemens console software. This plane was then used to position fMRI slices over the primary motor cortex.
None of the subjects received more than 200 effective TMS pulses per day, and half of them were at subthreshold intensity. The TMS protocol was in full accordance with the present safety guidelines for TMS (Rossi et al., 2009, Siebner et al., 2009).
Transcranial magnetic stimulation and electromyographic recordings Magnetic resonance imaging TMS was performed using a MagPro X100 stimulator and an MRcompatible TMS coil MRi-B91 (Magventure, Farum, Denmark). The stimulator generated biphasic pulses of approximately 280 μs duration. The TMS target region, the representational area of the right first dorsal interosseous (FDI) muscle on the primary motor cortex (M1) was defined for each subject based on the activation maps obtained from the finger tapping experiment during the first MRI session. After analysis of finger tapping data (SPM12), statistical parametric maps and anatomical scans were loaded into a neuronavigation suite (Brainsight 2, Rogue Research Inc., Canada). Subsequently subjects’ heads were registered to their respective anatomical scans using three reference points (nasion, left and right ear). The TMS coil attached to the MR coil array was registered using the manufacturer's calibration block and the coil tracker. Using the neuronavigation tool, positioning of the combined TMS/MR coil array within a distance of less than 5 mm to the target was achieved. Using the EMG integrated in the neuronavigation suite, the motor evoked response of the right FDI muscle was assessed on a grid of 13 points, one at the target (defined by the hotspot of activation maps from the finger tapping paradigm), and at 12 points around the hotspot on 2 rings, 4 points in the first and 8 in the second ring. The ring spacing was 5mm (Parameters for neuronavigation system: 3 rings, circular grid, 7 mm arc length, 5 mm ring spacing). The position with maximum TMS-evoked motor response in the FDI muscle was selected as the target to be used in the TMS/fMRI
Measurements were performed on a 3 T Tim Trio MR Scanner (Siemens Healthcare, Erlangen, Germany). During the first session (using the 32-channel head coil), finger-tapping data was acquired with a standard gradient-echo 2D single-shot EPI sequence with TR / TE = 1800 ms/38 ms, 23 slices, 3 mm slice thickness (1.8 mm slice gap), 1.5×1.5 mm2 in-plane resolution, and MA = 128×128. Anatomical MPRAGE images were also acquired (TR / TE / TI = 2300 ms / 4.21 ms/900 ms, flip angle 9°, 160 slices, 1.25 mm slice thickness, FOV = 240×56 mm2, MA = 240×256, and BW = 238 Hz/pixel). In the second session (using the dedicated MR coil array) functional images were acquired using the CMRR multi-band gradient-echo 2D single-shot EPI sequence (Moeller et al., 2010) with a limited number of slices to allow for higher temporal resolution (TR / TE = 1000 ms/ 33 ms, flip angle 60°, 14 slices, 3 mm slice thickness, 1.5×1.5 mm2 inplane resolution, MA = 128×128, BW = 1502 Hz/pixel and MB factor 2). Slices were aligned parallel to the TMS coil and fully covered the primary motor cortex. The MR coil array (Navarro de Lara, 2015) contained three MR-visible fiducial markers that were used to facilitate positioning the image slab based on a FLASH dataset (TR / TE = 400 ms/2.46 ms, flip angle 41°, 14 slices, 3 mm slice thickness, 0.69×0.69 mm2 in-plane resolution, MA = 256×320, GRAPPA = 2 and BW = 332 Hz/pixel). The three fiducial markers described above were identified in the FLASH images to define a plane parallel to the
Fig. 2. Measurement Protocol. (a) Flow diagram of the TMS experiment. (b) The stimulation protocol consists of 20 blocks, each block starting with 10 seconds baseline (fMRI acquisition but no TMS) followed by 10 seconds TMS (fMRI acquisition and TMS). The stimulation intensity of each block is chosen as one of the four intensities, i.e. 80%, 90%, 100% or 110% of individual's previously determined active motor threshold. (c) fMRI acquisition timing. The yellow blocks corresponds to the acquisition of one slab. fMRI data collection, blue line correspond to TMS stimulation pulse. Volumes are acquired with TR=1 s.
264
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
experiment. It corresponded well to the activations map hotspots. Motor thresholds were defined as the dI/dt values displayed by the stimulator that produced a MEP response between 0.05–0.1 mV within a time frame between 15 and 35 ms after application in the target muscle in two out of four trials during 50% of maximum voluntary contraction. Thresholds were determined outside of the MR environment, shortly before the scanning session. To avoid increase of cortical excitability during this procedure (Pascual-Leone et al., 1994) the stimulation intensity was gradually reduced until no muscle twitch was elicited in the target muscle in half of the trials. The motor threshold was verified in the MR scanner before running the stimulation protocol.
Combined TMS/fMRI The new MR coil is equipped with two sets of markers. The first set consists of three small, liquid-filled capsules that are mounted inside the MR coil's housing. These markers are visible on MR images and define a plane parallel to the MT/TMS coil used to position fMRI acquisition slices. The second set of markers consists of three small reflective spheres mounted on a non-ferromagnetic object and is part of the neuronavigation system. The latter markers are detachable and in this study were removed before moving the subject bed into the scanner. The TMS coil, rigidly combined with the MR coil array, was mounted to a robust coil-holding device (MRi Coil Holder, MagVenture, Farum, Denmark) and positioned using the neuronavigation suite on the redefined target in the scanner room but outside the scanner bore. Coil-orientation was tangential, approximately 45° to the mid-line. Subjects lay supine on an in-house made cushioned head holder, designed for comfortable subject fixation during the experiment. The TMS coil was connected using an extension cable of eight meters through a radiofrequency (RF) filter tube to a filter box mounted on the filter plate. This box also contained hardware to suppress leakage currents to values below 1 µA (Magventure, Farum, Denmark). Due to resistive cable losses and the filter transmission function, the TMS stimulator produces a lower dI/dt for a given stimulation intensity in the scanner, in comparison to the setup where the TMS coil is directly connected to the stimulator outside the scanner. For this reason, the motor thresholds determined outside the scanner had to be recorded as a dI/dt value rather than as percentage of the stimulator output. TMS coil position and the motor thresholds were briefly evaluated before the stimulation experiment by applying single TMS pulses. Head movement was restricted by foampadded cushions and the subject wore ear-plugs to reduce noise throughout the experiment. Acquiring one volume in the TMS/fMRI sequence took 680 ms (Fig. 2c). The minimum TMS delay times to avoid artefacts in the EPI images from TMS pulses were examined in a pilot study where we acquired MR images of a phantom using the same multiband gradientecho 2D single-shot EPI sequence as described above but varied the time after a TMS pulse of 100% stimulation output intensity to the next EPI acquisition. The mean of the first and last 20 image slabs (all without stimulation pulses) was used as a reference image. Twenty-six equidistantly spaced durations between 0 and 100 ms were applied as delay. This procedure was repeated five times. For each delay value, the root mean square (RMS) deviation from the reference image was calculated to obtain a measure of signal dropout. The results of this initial study are shown in Fig. 3c. It can be seen that a delay of 50 ms is sufficient to avoid TMS effects on image quality. In the actual experiment, a much longer delay of 160 ms was chosen, so to place TMS pulses right in the middle of non-acquisition periods (repetition time TR=1000 ms, acquisition time TA=680 ms, time between acquisition TD=320 ms; see also Fig. 2).
Fig. 3. Functional images and determination of minimal TMS delay. Individual functional images acquired with the novel coil during (a) a rest block, and (b) a stimulation block. Slices were acquired parallel to the TMS coil plane. Green arrows indicate the interhemispheric fissure. (c) Percentage of the root mean square error versus time after TMS pulse to the next image acquisition. It can be seen that a minimum delay of 50 ms is sufficient to avoid artifacts or dropouts in the images. For our experiment, we chose a delay time of 160 ms (marked with the blue flash).
Data analysis Data analysis was performed using SPM12b. Pre-processing of fMRI acquisitions included slice-timing correction (Sladky et al., 2011), motion correction and smoothing with a 3 mm FWHM isotropic Gaussian kernel. The pre-processed datasets were analysed using the general lineal model (GLM) as implemented in SPM, i.e. linear regression was performed at each voxel, using generalized least squares with a global approximate AR(1) autocorrelation model. Based on the TMS stimulation intensities (TMS experiment) or movement epochs (finger tapping experiment), models were defined and convolved with SPM's haemodynamic response function. Activation maps for single subjects were obtained by applying a t-test to the GLM parameter estimates obtained for each stimulation intensity or movement epoch. The significance level was set at p < 0.05, family-wise error (FWE) corrected at the voxel level. For visualisation purposes, anatomical images acquired with the 32-channel head coil were co-registered to the anatomical images acquired with the 7-channel MR coil array with an in-house written software in MATLAB, which applied N4 bias correction (Tustison et al., 2010) prior to performing rigid registration using SPM. In order to calculate the coordinates of the point where the central axis of the TMS coil penetrated the cortex, a skull-stripped version of the co-registered anatomical data (acquired with the 32-channel head coil) using the AFNI tool 3dSkullStrip was generated. Then, the intersection of brain surface and TMS coil axis as obtained from three 265
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Single subject results
markers placed in the co-planar MR coil was calculated with Matlab code written in-house. These subject-space coordinates were transformed to MNI space by using the unified segmentation routine as implemented in SPM12 applied to the co-registered 32-channel anatomical data. For each subject, three voxels were defined, all based on the activation maps from the 110% stimulation amplitude blocks: (i) global maximum voxel (activation maximum over the whole brain); (ii) Finger-Tapping-M1 maximum voxel (activation maximum within the M1 hand area as obtained from the individual's finger tapping activation map); (iii) MNI-M1 maximum voxel (activation maximum within the M1 hand area as defined by transforming the hand knob MNI coordinate [−40 −20 52] (Cardenas-Morales et al., 2014) to subject space). For (ii), a search radius of 2 cm was applied to determine the voxel of maximum M1 activation. For (iii), the “find local maximum” functionality as implemented in SPM12 was used, starting from MNI coordinate [−40 −20 52]. BOLD signal intensity changes for each subject were calculated from the parameter estimates obtained by SPM divided by the mean signal (constant term in the design matrix). Mean signal time courses for finger tapping and all stimulation intensities were calculated as a percentage with respect to the last two time points of each experimental block (Bestmann et al., 2004).
Activation maps of a typical subject are shown in Fig. 4. Maps show activation from the finger tapping experiment (Fig. 4a) and during TMS with 110% (Fig. 4b, d) and 100% aMT stimulation levels (Fig. 4c, e). Each map is centred on the respective maximum, i.e. maximum finger tapping activation (Fig. 4a), global maximum (Fig. 4b, c), and M1 maximum (Fig. 4d, e). All maps are thresholded at p < 0.05 (voxel-level FWE-corrected). Table 1 shows the activation t-values and the parameter estimates at the three extracted voxels for all subjects and for all stimulation intensities. The FWE corrected and uncorrected p-values are also listed in the table. For the subject shown in Fig. 4, global activation maxima were statistically significant (p < 0.05, FWE-corrected) for all TMS intensities, while M1 maxima were found significant only for 100% and 110% motor threshold stimulation intensity (see Table 1, Subject 4). For all subjects, global maximal activations were found below the coil center.
Group results To evaluate coil sensitivity, we focused on the quality of the acquired signal in terms of BOLD signal intensity change and statistical significance. For this purpose, group-level results were calculated as the mean of the percentage BOLD signal intensity change for the three voxels defined above. The results are plotted in Fig. 5 for each stimulation intensity as mean and standard error of the mean (SEM) values. Across the group, signal changes (mean ± SEM, [%]) at the global maximum were 0.88 ± 0.24, 1.09 ± 0.29, 1.65 ± 0.43, 2.77 ± 0.73 for 80%, 90%, 100%, and 110%, respectively. Corresponding values in local M1 maxima based on the finger tapping results were 0.43 ± 0.30, 0.63 ± 0.30, 1.01 ± 0.23, and 2.01 ± 0.43, respectively. For MNI-based maxima signal changes were calculated as 0.73 ± 0.28, 0.91 ± 0.37, 1.34 ± 0.49 and 2.21 ± 0.61, respectively. Across the group, signal changes in the global maximum were statistically significant for all stimulation intensities. At the voxel of M1 maxima only values for the 100% and 110% stimulation intensities were statistically significant across subjects. In both graphs, an ascending monotonic relationship of mean signal changes over stimulation intensity can be seen.
Results None of the subjects reported any adverse side effects after participation. Functional images acquired with the new dedicated MR coil array are shown in Fig. 3a and b in both states, during stimulation and rest blocks. None of the volumes acquired during the stimulation protocol were affected by TMS pulses or presented dropouts under the coil. From the measurements in the phantom, we concluded that ensuring a wait time of at least 50 ms before EPI acquisition after a TMS pulse, no distortions or artefacts on the functional images were produced either by the magnetic field of the TMS pulses or possible vibrations (see Fig. 3c). No susceptibility effects due to the presence of the TMS system were observed.
Fig. 4. Single subject results (subject 4). Arrows point to the activation peak used for extracting parameters estimates and time courses. (a) Activation map obtained from the finger tapping experiment. (b)-(e) TMS-induced activation for 110% and 100% aMT stimulation at the global maximum (b-c) and the local maximum next to the M1 area as defined via the finger tapping results (d-e).
266
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Table 1 Peak activation t-values, parameter estimates, voxel-wise FWE-corrected and uncorrected p-values at the three extracted voxels, i.e., the global activation maximum, the local activation maximum near M1 based on the individual finger tapping localizer, and the local activation maximum near M1 based on fixed MNI coordinates, respectively. Global Activation Maximum
Subject 1
Subject 2
Subject 3
Subject 4
Subject 5
Subject 6
Subject 7
Levels
t-Value
beta
p corr. (FWE)
p unc.
Local Activation Maximum near M1 based on Finger Tapping t-Value beta p corr. (FWE) p unc.
110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80 110 100 90 80
15.36 9.11 3.97 7.43 8.92 6.67 5.14 3.71 10.38 8.83 10.26 8.87 19.24 15.58 10.83 8.18 19.87 10.97 2.52 0.49 11.59 6.24 5.92 4.82 15.32 3.69 3.57 1.45
11.2 6.51 3.16 5.02 11.15 7.94 5.75 5.72 1.84 1.55 1.67 1.61 21.19 16.1 12.17 8.92 17.23 8.97 2.26 0.42 6.85 3.6 3.39 2.96 7.51 1.88 1.92 0.79
< 10-10 < 10-10 1 < 10-7 < 10-10 < 10-5 0.048 1 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-9 < 10-10 < 10-10 1 1 < 10-10 < 10-3 < 10-3 0.24 < 10-10 1 1 1
< 10-16 < 10-16 < 10-4 < 10-12 < 10-16 < 10-10 < 10-6 < 10-3 < 10-16 < 10-14 < 10-16 < 10-16 < 10-16 < 10-16 < 10-16 < 10-4 < 10-16 < 10-16 0.0061 0.31 < 10-16 < 10-9 < 10-8 < 10-5 < 10-16 < 10-3 < 10-3 0.074
15.36 9.11 3.97 7.43 7.32 5.02 3.33 2.04 10.38 8.83 10.26 8.87 16 8.25 5.11 -0.21 15.54 5.85 -2.66 -4.15 11.59 6.24 5.92 4.82 12.04 1.2 0.67 0.37
< 10-10 < 10-10 1 < 10-7 < 10-6 0.087 1 1 < 10-10 < 10-10 < 10-10 < 10-10 < 10-10 < 10-9 0.057 1 < 10-10 0.0012 1 1 < 10-10 < 10-3 < 10-3 0.24 < 10-10 1 1 1
11.2 6.51 3.16 5.02 5.93 3.87 2.41 2.04 1.84 1.55 1.67 1.61 12.06 5.83 3.93 -0.15 10.65 3.78 -1.88 -2.85 6.85 3.6 3.39 2.96 5 0.51 0.31 0.16
< 10-16 < 10-16 < 10-4 < 10-12 < 10-12 < 10-6 < 10-3 0.021 < 10-16 < 10-14 < 10-16 < 10-16 < 10-16 < 10-8 < 10-4 0.58 < 10-16 < 10-8 1 1 < 10-16 < 10-9 < 10-8 < 10-5 < 10-16 0.12 0.25 0.36
Local Activation Maximum near M1 based on MNI t-Value beta p corr. p unc. (FWE) 10.08 2.89 0.85 2.28 8.92 6.67 5.14 3.71 6.88 2.46 5.78 4.99 16 8.25 5.11 -0.21 6.85 2.32 -0.99 -0.48 6.26 4.32 1.40 4.15 6.76 0.04 -0.60 -1.08
6.25 1.76 0.58 1.31 11.15 7.94 5.75 5.72 3.09 1.1 2.38 2.29 12.06 5.83 3.93 -0.15 7.27 2.32 -1.08 -0.51 3.11 2.09 0.68 2.14 2.46 0.02 -0.24 -0.44
< 10-10 1 1 1 < 10-10 < 10-5 0.048 1 < 10-5 1 0.0017 0.1 < 10-10 < 10-9 0.057 1 < 10-5 1 1 1 < 10-3 0.98 1 1 < 10-5 1 1 1
< 10-16 0.002 0.2 0.01 < 10-16 < 10-10 < 10-6 < 10-3 < 10-10 0.0072 < 10-8 < 10-6 < 10-16 < 10-8 < 10-4 0.58 < 10-10 0.01 0.84 0.68 < 10-9 < 10-5 0.081 < 10-4 < 10-10 0.48 0.73 0.86
Note that the FWE-correction implementation used limited p-value accuracy to < 10–10.
Fig. 5. Group results for M1 after TMS stimulation. Mean BOLD signal change and SEM over the whole group (n=7) across the different stimulation intensities in the voxel of (a) global maximal activation, (b) local maximum activation near M1 based on the FT results, and c) local maximum activation near M1 based on the MNI coordinates. Results marked with an asterisk are statistically significant (p < 0.05). Error bars indicate standard errors of the means. It can be seen that activation increases with stimulation amplitude in all regions examined. At the voxel of global maximum activation, activation was statistically significant for all stimulation intensities. Local maxima were significant for 100% and 110% aMT stimulation only.
caused signal increase of 1.3% (1.0%), 0.8% (0.5%) and 0.6% (0.3%) for global, finger tapping and MNI maxima, respectively. The presented values are slightly different than the values calculated using the parameter estimates as obtained with SPM. This difference can be explained by the different approaches used for the baseline signal calculation. Mean MNI coordinates and spread for the reported peaks were calculated together with the projection of the TMS coil centre onto the brain. The spread was defined as the mean of the standard deviation in
In order to examine TMS-induced activation changes in more detail, time courses were extracted and averaged over repeated blocks of the same intensity and then averaged across subjects. Fig. 6 shows the resulting plots for the global maximum (a) and the local maxima based on finger tapping (b) and MNI-coordinate (c). Time courses for the finger tapping runs are displayed for comparison. For 110% stimulation intensity we obtain 3.3%, 2.4%, 1.5% for global, finger tapping and MNI maxima, respectively. 100% stimulation resulted in 2.1%, 1.4% and 0.9%. Subthreshold stimulation with 90% (80%)
267
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
primary motor cortex using four different stimulation intensities. Our results show that observed BOLD responses to TMS are considerably higher compared to previous publications where the standard birdcage coil setup was used (Baudewig et al., 2001; Bestmann et al., 2004). It seems likely that this is due to the greatly improved sensitivity due to the new setup. A monotonically ascending relationship of BOLD signal change with TMS intensity has been obtained in all three regions examined, i.e. global maximum voxel, Finger-Tapping-M1 maximum voxel, MNI-M1 maximum voxel. Only the global maximum voxel showed statistically significant signal changes for all stimulation intensities. Consistent with results obtained in other studies (Bohning et al., 1999, Baudewig et al., 2001, Bestmann et al., 2003b, 2004), the results on the local maxima next to M1 were found to be significant for the 100% and 110% stimulation intensities. At this particular M1 site, earlier physiological studies (Fisher et al., 2002) proposed that intensities under motor threshold affect low-threshold inhibitory nodes, which produce inhibitory synaptic activity within the primary motor cortex (Bestmann et al., 2003b). A sigmoidal effect between TMS intensities and activation directly at M1 has been proposed based on these effects (Bestmann et al., 2003b, Hanakawa et al., 2009). It is important to note that no susceptibility effects or distortions of the images were observed beneath the coil in the fMRI data. We suggest that BOLD changes detected below the coil, i.e., the stimulation site, reflect an intensity-dependent increase in local neuronal activity even below the individual motor threshold. On the other hand, haemodynamic response at M1 was observed at threshold levels and above, while responses to subthreshold stimulation were below significance level, possibly due to compensatory effects caused by subcomponents of the motor network. Fig. 6 shows that the acquired signal returns to baseline within the fixed 10 s rest period of our stimulation protocol. Carry-over effects may still have occurred and could explain the variance seen for low intensities. In this study, the 2 out of 4 rule was used to determine the motor threshold. This approach differs from the established 5 out of 10 rule, and might have caused a slight misestimation of threshold levels. The results obtained in this study also demonstrate the applicability of the novel setup to acquire high-quality functional images during stimulation. The time courses presented are now comparable to BOLD signal changes obtained in typical fMRI block-design experiments, with peaks of about 3% signal change. A limitation of previous TMS/fMRI setups concerned TMS coil positioning. In the new setup, positioning the TMS coil is greatly facilitated, as mounting the TMS coil is no longer confined to the limited space within a birdcage coil, enhancing flexibility. Also, we have shown here that neuronavigation tools can now be directly applied, making this cumbersome task easier and more accurate. It also opens up the possibility to perform concurrent TMS/fMRI studies in new target regions with no immediate sensory output. In addition to mapping local TMS-induced activation, one of the main goals of neuroimaging studies combined with brain stimulations techniques is to investigate the distributed effects it can have on functional brain networks (Bestmann and Feredoes, 2013). For this reason the assessment of remote effects of the stimulation represents an important aspect that has to be fulfilled by the neuroimaging technique. The proposed set up with one 7-channel surface coil array is limited in this way. However, in order to answer these more complex research questions, such as TMS-remote effects, we propose the use of an additional MR coil array positioned opposite to the site of stimulation. Using the combination of the two MR coil arrays with advanced fast and precise imaging and neuronavigation methods, our group has already demonstrated acute local and network effects of TMS over the left DLPFC (Tik et al., 2016). Using the newly developed hardware comprising seven RF receive coil elements, tightly fixed to the TMS coil, it is now possible to benefit from parallel and multiband imaging approaches, thereby improving
Fig. 6. : Group-averaged BOLD signal time courses averaged over blocks Stimulation blocks are indicated in green, and time courses extracted from finger tapping data are shown as dashed lines. Colors correspond to stimulation amplitudes: 80% (lilac), 90% (yellow), 100% (orange), 110% (blue). Data is plotted for the global maximum (a) and the local M1 maxima based on finger tapping (b) and MNI-coordinate (c). Table 2 Mean MNI coordinates and mean of the standard deviation given as the spread. These values were calculated for the reported peaks and for the voxel found as the projection of the TMS coil center on the brain. Voxel Specification
MNI Coordinates [x, y, z] Mean
Standard Deviation
TMS coil center projection on brain surface
[−40, −15, 64]
[9.3, 11.9, 7.4]
Voxel at global activation maximum
[−40, −17, 59] [−42, −24, 55] [−38, −15, 54]
[6.2, 12.4, 8.1]
Voxel at local activation maximum near M1 (based on finger tapping) Voxel at local activation maximum near M1 (MNI based)
[4.3, 10.7, 4.1] [3.1, 4.5, 3.6]
each direction. The respective coordinates are summarised in Table 2. Discussion Our results clearly demonstrate that the new setup using a novel, dedicated MR coil array can be successfully applied to conduct concurrent high sensitivity TMS/fMRI experiments with high temporal and spatial resolution. The 7-channel MR receive array has already been shown to boost sensitivity compared to the standard birdcage coil approach (Navarro de Lara et al., 2015). In this study we show how this benefit in sensitivity can be used to assess TMS-induced activity in the 268
NeuroImage 150 (2017) 262–269
L.I. Navarro de Lara et al.
Denslow, S., Lomarev, M., George, M.S., Bohning, D.E., 2005. Cortical and subcortical brain effects of transcranial magnetic stimulation (TMS)-induced movement: an interleaved TMS/functional magnetic resonance imaging study. Biol. Psychiatry 57, 752–760. Fisher, R.J., Nakamura, Y., Bestmann, S., Rothwell, J.C., Bostock, H., 2002. Two phases of intracortical inhibition revealed by transcranial magnetic threshold tracking. Exp. Brain Res. 143, 240–248. Hanakawa, T., Mima, T., Matsumoto, R., Abe, M., Inouchi, M., Urayama, S., Anami, K., Honda, M., Fukuyama, H., 2009. Stimulus-response profile during single-pulse transcranial magnetic stimulation to the primary motor cortex. Cereb. Cortex 19, 2605–2615. Lefaucheur, J.P., Andre-Obadia, N., Antal, A., Ayache, S.S., Baeken, C., Benninger, D.H., Cantello, R.M., Cincotta, M., de Carvalho, M., De Ridder, D., Devanne, H., Di Lazzaro, V., Filipovic, S.R., Hummel, F.C., Jaaskelainen, S.K., Kimiskidis, V.K., Koch, G., Langguth, B., Nyffeler, T., Oliviero, A., Padberg, F., Poulet, E., Rossi, S., Rossini, P.M., Rothwell, J.C., Schonfeldt-Lecuona, C., Siebner, H.R., Slotema, C.W., Stagg, C.J., Valls-Sole, J., Ziemann, U., Paulus, W., Garcia-Larrea, L., 2014. Evidence-based guidelines on the therapeutic use of repetitive transcranial magnetic stimulation (rTMS). Clin. Neurophysiol. 125, 2150–2206. Moeller, S., Yacoub, E., Olman, C.A., Auerbach, E., Strupp, J., Harel, N., Ugurbil, K., 2010. Multiband multislice GE-EPI at 7 T, with 16-fold acceleration using partial parallel imaging with application to high spatial and temporal whole-brain fMRI. Magn. Reson. Med. 63, 1144–1153. Moisa, M., Siebner, H.R., Pohmann, R., Thielscher, A., 2012. Uncovering a contextspecific connectional fingerprint of human dorsal premotor cortex. J. Neurosci.: Off. J. Soc. Neurosci. 32, 7244–7252. Navarro de Lara, L.I., Windischberger, C., Kuehne, A., Woletz, M., Sieg, J., Bestmann, S., Weiskopf, N., Strasser, B., Moser, E., Laistler, E., 2015. A novel coil array for combined TMS/fMRI experiments at 3 T. Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med. / Soc. Magn. Reson. Med. 74, 1492–1501. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia 9, 97–113. Pascual-Leone, A., Valls-Sole, J., Wassermann, E.M., Hallett, M., 1994. Responses to rapid-rate transcranial magnetic stimulation of the human motor cortex. Brain: J. Neurol. 117 (Pt 4), 847–858. Rossi, S., Hallett, M., Rossini, P.M., Pascual-Leone, A., Grp, S.T.C., 2009. Safety, ethical considerations, and application guidelines for the use of transcranial magnetic stimulation in clinical practice and research. Clin. Neurophysiol. 120, 2008–2039. Rossini, P., Rossi, S., 2007. Transcranial magnetic stimulation: diagnostic, therapeutic, and research potential. Neurology 68, 484–488. Ruff, C.C., Bestmann, S., Blankenburg, F., Bjoertomt, O., Josephs, O., Weiskopf, N., Deichmann, R., Driver, J., 2008. Distinct causal influences of parietal versus frontal areas on human visual cortex: evidence from concurrent TMS-fMRI. Cereb. Cortex 18, 817–827. Ruff, C.C., Blankenburg, F., Bjoertomt, O., Bestmann, S., Freeman, E., Haynes, J.D., Rees, G., Josephs, O., Deichmann, R., Driver, J., 2006. Concurrent TMS-fMRI and psychophysics reveal frontal influences on human retinotopic visual cortex. Curr. Biol.: CB 16, 1479–1488. Ruff, C.C., Blankenburg, F., Bjoertomt, O., Bestmann, S., Weiskopf, N., Driver, J., 2009. Hemispheric differences in frontal and parietal influences on human occipital cortex: direct confirmation with concurrent TMS-fMRI. J. Cogn. Neurosci. 21, 1146–1161. Shitara, H., Shinozaki, T., Takagishi, K., Honda, M., Hanakawa, T., 2011. Time course and spatial distribution of fMRI signal changes during single-pulse transcranial magnetic stimulation to the primary motor cortex. NeuroImage 56, 1469–1479. Siebner, H.R., Bergmann, T.O., Bestmann, S., Massimini, M., Johansen-Berg, H., Mochizuki, H., Bohning, D.E., Boorman, E.D., Groppa, S., Miniussi, C., PascualLeone, A., Huber, R., Taylor, P.C., Ilmoniemi, R.J., De Gennaro, L., Strafella, A.P., Kahkonen, S., Kloppel, S., Frisoni, G.B., George, M.S., Hallett, M., Brandt, S.A., Rushworth, M.F., Ziemann, U., Rothwell, J.C., Ward, N., Cohen, L.G., Baudewig, J., Paus, T., Ugawa, Y., Rossini, P.M., 2009. Consensus paper: combining transcranial stimulation with neuroimaging. Brain Stimul. 2, 58–80. Silvanto, J., Pascual-Leone, A., 2012. Why the assessment of causality in brain-behavior relations requires brain stimulation. J. Cogn. Neurosci. 24, 775–777. Sladky, R., Friston, K.J., Trostl, J., Cunnington, R., Moser, E., Windischberger, C., 2011. Slice-timing effects and their correction in functional MRI. Neuroimage 58, 588–594. Tik M., Woletz M., Navarro de Lara L., Sladky R., Hoffmann A., Hummer A., Windischberger C., 2016. Multimodal Assessment of Local and Remote TMSinduced Effects. In: Annual Meeting of the Organization for Human Brain Mapping Geneva, Switzerland. Tustison, N.J., Avants, B.B., Cook, P.A., Zheng, Y., Egan, A., Yushkevich, P.A., Gee, J.C., 2010. N4ITK: improved N3 bias correction. IEEE Trans. Med. Imag. 29, 1310–1320. Walsh, V., Pascual-Leone, A., 2005. Transcranial Magnetic Stimulation: a Neurochronometrics of Mind. MIT Press, Cambridge, MA. Weiskopf, N., Josephs, O., Ruff, C.C., Blankenburg, F., Featherstone, E., Thomas, A., Bestmann, S., Driver, J., Deichmann, R., 2009. Image artifacts in concurrent transcranial magnetic stimulation (TMS) and fMRI caused by leakage currents: modeling and compensation. J. Magn. Reson. Imag.: JMRI 29, 1211–1217. Wiggins, G.C., Triantafyllou, C., Potthast, A., Reykowski, A., Nittka, M., Wald, L.L., 2006. 32-channel 3 T receive-only phased-array head coil with soccer-ball element geometry. Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med. / Soc. Magn. Reson. Med. 56, 216–223.
temporal and spatial resolution. Until now, these new acquisition techniques, which have become standard in current fMRI experiments, were not available to concurrent TMS/fMRI studies. The improvement in sensitivity, which in turn can yield higher spatial resolution, together with better temporal resolution by applying parallel imaging techniques may help to improve the specificity of concurrent TMS/fMRI experiments. Taken together, here we show the applicability of neuronavigated concurrent TMS/fMRI using a new dedicated high-sensitivity coil array for precise mapping of local intensity-dependent BOLD responses even at an individual subject's level. This flexible setup will allow for future experiments targeting different cortical targets to understand TMS mechanism of action and improve therapeutic effects of TMS in clinical settings. Acknowledgements We thank Gunnar Hallsson, Claus Mathiesen and colleagues from MagVenture (Denmark) for their support in performing this study. This work has been financially supported by the HRSM Austrian BMWFJ FFG Project Nr. 832107, “Vienna Research Studio for Ultra-High Field Magnetic Resonance Applications”. References Auerbach, E.J., Xu, J., Yacoub, E., Moeller, S., Ugurbil, K., 2013. Multiband accelerated spin-echo echo planar imaging with reduced peak RF power using time-shifted RF pulses. Magn. Reson. Med.: Off. J. Soc. Magn. Reson. Med. / Soc. Magn. Reson. Med. 69, 1261–1267. Baudewig, J., Siebner, H., Bestmann, S., Tergau, F., Tings, T., Paulus, W., Frahm, J., 2001. Functional MRI of cortical activations induced by transcranial magnetic stimulation (TMS). Neuroreport 12, 3543–3548. Bestmann, S., Baudewig, J., Frahm, J., 2003a. On the synchronization of transcranial magnetic stimulation and functional echo-planar imaging. J. Magn. Reson. imaging: JMRI 17, 309–316. Bestmann, S., Baudewig, J., Siebner, H.R., Rothwell, J.C., Frahm, J., 2003b. Subthreshold high-frequency TMS of human primary motor cortex modulates interconnected frontal motor areas as detected by interleaved fMRI-TMS. NeuroImage 20, 1685–1696. Bestmann, S., Baudewig, J., Siebner, H.R., Rothwell, J.C., Frahm, J., 2004. Functional MRI of the immediate impact of transcranial magnetic stimulation on cortical and subcortical motor circuits. Eur. J. Neurosci. 19, 1950–1962. Bestmann S., Feredoes E., 2013. Combined neurostimulation and neuroimaging in cognitive neuroscience: past, present, and future. Annals of the New York Academy of Sciences 1296, pp. 11–30. 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 with concurrent TMS-fMRI. Neuropsychologia 44, 2959–2971. Bestmann, S., Swayne, O., Blankenburg, F., Ruff, C.C., Haggard, P., Weiskopf, N., Josephs, O., Driver, J., Rothwell, J.C., Ward, N.S., 2008. Dorsal premotor cortex exerts state-dependent causal influences on activity in contralateral primary motor and dorsal premotor cortex. Cereb. Cortex 18, 1281–1291. Bohning, D., Shastri, A., McConell, K., Nahas, Z., Lorberbaum, J., Roberts, D., Teneback, C., Vincent, D., George, M., 1999. A combined TMS/fMRI Study of intensitydependent TMS over motor cortex. Biol. Psychiatry 45, 385–394. Bohning, D., Shastri, A., Nahas, Z., JP, L., Andersen, S., Dannels, W., Haxthausen, E., Vincent, D., George, M., 1998. Echoplanar BOLD fMRI of brain activation induced by concurrent transcranial magnetic stimulation. Investig. Radiol. 33, 336–340. Bohning, D.E., Shastri, A., Lomarev, M.P., Lorberbaum, J.P., Nahas, Z., George, M.S., 2003. BOLD-fMRI response vs. transcranial magnetic stimulation (TMS) pulse-train length: testing for linearity. J. Magn. Reson Imag. 17, 279–290. Bohning, D.E., Shastri, A., Wassermann, E.M., Ziemann, U., Lorberbaum, J.P., Nahas, Z., Lomarev, M.P., George, M.S., 2000. BOLD-f MRI response to single-pulse transcranial magnetic stimulation (TMS). J. Magn. Reson. Imag.: JMRI 11, 569–574. Bungert, A., Chambers, C.D., Phillips, M., Evans, C.J., 2012. Reducing image artefacts in concurrent TMS/fMRI by passive shimming. NeuroImage 59, 2167–2174. Cardenas-Morales, L., Volz, L.J., Michely, J., Rehme, A.K., Pool, E.M., Nettekoven, C., Eickhoff, S.B., Fink, G.R., Grefkes, C., 2014. Network connectivity and individual responses to brain stimulation in the human motor system. Cereb. Cortex 24, 1697–1707. Chen, R., Cros, D., Curra, A., Di Lazzaro, V., Lefaucheur, J., Magistris MaZ, U., 2008. The clinical diagnostic utility of transcranial magnetic stimulation: report of an IFCN committee. Clin. Neurophysiol.: Off. J. Int. Fed. Clin. Neurophysiol. 119 (3), 504–532.
269