PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Hum Brain Mapp. Author manuscript; available in PMC 2010 June 29.
Published in final edited form as:
PMCID: PMC2893589
NIHMSID: NIHMS211459

Motor Threshold in Transcranial Magnetic Stimulation: The Impact of White Matter Fiber Orientation and Skull-to-Cortex Distance

Abstract

The electrophysiology of transcranial magnetic stimulation (TMS) of motor cortex is not well understood. In this study, we investigate several structural parameters of the corticospinal tract and their relation to the TMS motor threshold (MT) in 17 subjects, with and without schizophrenia. We obtained structural and diffusion tensor MRI scans and measured the fractional anisotropy and principal diffusion direction for regions of interest in the corticospinal tract. We also measured the skull-to-cortex distance over the left motor region. The anterior–posterior trajectory of principle diffusion direction of the corticospinal tract and skull-to-cortex distance were both found to be highly correlated with MT, while fractional anisotropy, age and schizophrenia status were not. Two parameters—skull-to-cortex distance and the anterior component of the principle diffusion direction of the corticospinal tract as it passes the internal capsule—are highly predictive of MT in a linear regression model, and account for 82% of the variance observed (R2 = 0.82, F = 20.27, P < 0.0001) in measurements of MT. The corticospinal tract’s anterior–posterior direction alone contributes 13% of the variance explained.

Keywords: TMS, DTI, motor threshold, fiber-tracking, MRI

INTRODUCTION

Transcranial magnetic stimulation (TMS) has become a tool of significant importance in both basic and clinical neurosciences [George and Belmaker, 2000]. The basic neurophysiology of TMS is incompletely understood with the exact relationship between TMS coil orientation, electric field strength, neuron orientation and the generation of transmembrane excitation still subject to conjecture [George and Belmaker, 2000]. One very practical consequence of this interaction between induced electrical field and neuronal excitation is the motor threshold (MT). MT is defined as the minimum TMS intensity sufficient to produce a predefined motor-evoked potential (MEP) in the contralateral abductor pollicis brevis in at least 50% of trials [Rossini et al., 1994]. MT, roughly a measure of the TMS intensity necessary to evoke a peripheral motor response, is highly variable across individuals but also remarkably constant in a given individual [Cicinelli et al., 1997; Mills and Nithi, 1997; Ziemann et al., 1996]. The MT is of particular importance in TMS experimental design as a way of calibrating and normalizing TMS coil output energy for individual physiologic variability and thus determines both dose and safety limits. As TMS is adapted for therapeutic applications, the MT will likely assume importance for calculation of patient-specific therapeutic TMS dosage. As with TMS physiology, the underlying variables explaining the between-subject variance in MT are unclear. Thus, both far age [Kozel et al., 2000] and skull cortex distance (SCD) [Kozel et al., 2000; McConnell et al., 2001; Stokes et al., 2005] (likely correlated to age) have been shown to increase linearly with MT.

Diffusion tensor imaging (DTI) is a well established magnetic resonance imaging technique used to quantify diffusivity of water in tissue [Le Bihan et al., 1986]. In the brain, water diffusion is constrained by microscopic structures including myelin sheaths around axons. Because of this, DTI has been useful for studying white matter anatomy and pathology in the central nervous system [Kubicki et al., 2007; Taylor et al., 2004; Wakana et al., 2004]. DTI can image the course of large white matter structures—so called “fiber-tracking.”

Our a priori hypothesis was that properties of the corticospinal tract involved in transmitting the TMS MEP are related to MT. We used data from a study exploring cortical excitability in unmedicated schizophrenics and healthy controls in an attempt to find a generalizable, pathology-independent, physiological result. We used DTI to explore white matter correlates of MT and MEP conduction. In this study, we investigate DTI-derived measures of corticospinal tract orientation and white matter integrity. We also measure the SCD over the left motor cortex. We propose a new method for simply and reliably defining an informative ROI within the corticospinal tract using DTI fiber-tracking techniques. Finally, we present a linear regression model that accounts for much of the variability in MT as a function of SCD over motor cortex fiber orientation in the corticospinal tract.

METHODS

Subjects

Subjects were enrolled as part of a comprehensive study investigating cortical excitability in unmedicated schizophrenia patients and healthy controls, matched for age, gender, education, race and smoking status. The study was funded by the NIMH (R21-MH065630-01A1) and approved by Medical University of South Carolina Institutional Review Board. While most of the patients were recruited from local community mental health centers, a local homeless shelter, and referrals from families and private practitioners, almost all the control nonschizophrenia subjects were recruited through local advertizing and outreach. An average of two people was screened per week over a period of 18 months (from August 2004 until April 2006, ~150). All eligible patients completed the MacArthur Competency Assessment Tool—Clinical Research (MacCAT-CR) [Berg et al., 1996] to insure validity of the informed consent process. After demonstration of competence, the subjects completed a safety screen to undergo TMS and MRI scanning. Of those screened, a total of 43 participants signed informed consent; 20 met all criteria for this study and completed all needed assessments. A total of 17 had all the necessary measurement: MT, scalp cortex distance and DTI scans. Twelve had a primary diagnosis of schizophrenia (DSM-IV) as determined by the structured clinical interview diagnosis (SCID). All but five were right-handed. All subjects underwent an evaluation visit, a TMS session and a scanning session within less than 1 week apart. For this report, all tested negatively for illicit drugs or psychotropic medications and refrained from nicotine and caffeine for 2 h before EMG assessments.

Magnetic Stimulation of the Brain

After a brief description of the experiment and demonstration of TMS, subjects inserted ear plugs and were seated; the experimental apparatus was adjusted to a height to allow each subject to comfortably place their chin in a chin rest. The EMG was recorded using two pregelled Nicolet 20 × 25 mm2 Ag–AgCl disposable electrodes; these electrodes were placed over the region of the abductor pollicus brevis (APB) belly and associated tendon of the right hand. Additionally, a 40 × 50 mm2 pregelled Ag–AgCl ground electrode was placed on the back of the right hand. EMG activity was digitized at 5 kHz with a bandpass between 0.5 and 1,000 Hz, and was filtered at 60 Hz, using the Micro 1401 MK II and CED 1902 signal conditioner (Cambridge Electronic Design). The signal was further filtered with a high-pass filter with corner frequency of 30 Hz and transition gap of 13 Hz. The output of these filtering stages was manually examined to confirm that the signal of interest was in no way significantly distorted.

The TMS coil consisted of a figure-of-eight magnet (mean diameter of each lobe = 8 cm) and powered by a Magstim 200+ stimulator (Magstim, Whitland, Dyfed, UK). Its center was initially positioned 5 cm lateral to the vertex on the interauricular line with the handle in line with the parasagittal plane. TMS coil placement was then optimized by delivering pulses and examining the resulting MEP displayed through Spike2 software. Resting MT (rMT) was defined using a modified best parameter estimation by sequential testing (BEST-PEST) method [Awiszus, 2003] that dynamically adjusts TMS intensity based on a minimum MEP amplitude of 50 μV elicited at a given stimulation intensity [Mishory et al., 2004]. rMT was somewhat higher than previously reported values [Harris-Love et al., 2007] (healthy controls: mean = 72% of maximum stimulator output, SD = 11%; schizophrenia: mean = 63%, SD = 16%) and did not differ significantly between the groups (P = 0.19). Coil orientation was such that the line formed by the intersection of the two figure-of-eight lobes was aligned with the A–P axis. Although this is done in an ad-hoc fashion, our internal lab data show there to be a less than 5 degree variation from the A–P axis in either direction.

Magnetic Resonance Imaging Protocol

MRI scanning was performed using a 3 T clinical MRI scanner with a SENSE coil (Intera, Philips Medical Systems, Bothell, WA). A survey scan ascertained head location for subsequent anatomical and DTI scans. The T1-weighted structural scan was a rapid 3D gradient echo scan. It was acquired sagittally, covering the whole brain and using the following parameters: TR (for each FLASH line) = 11.2 ms, TE = 5.7 ms, slice thickness = 1 mm, TI = 960 ms, TR (for each magnetization preparation loop) = 3,000 ms, field of view (FOV) = 256 mm, number of slices = 160, matrix = 256 × 256. DTI acquisition consisted of a single-shot spin echo, echo planar imaging acquisition with a parallel imaging factor of 2 and partial Fourier factor of 80%, and the following parameters, TR = 5 s, TE = 100 ms, slice thickness = 3 mm, gap = 0.3 mm, FOV = 256 mm, number of slices = 30, matrix = 112 × 112 (interpolated to 256 × 256). For each slice, one image was acquired with no additional diffusion weighting and six diffusion-weighted images with b = 1,000 s/mm2 using the scheme described elsewhere [Basser and Pierpaoli, 1998]. The diffusion-weighted acquisitions were repeated three times and averaged on the scanner to improve the signal-to-noise ratio. The resulting total acquisition time was 110 s.

Image Processing Cortical Distance Measurement

A Brodmann area model (provided as part of the MRI-Cro package, http://www.sph.sc.edu/comd/rorden/mricro.html) was transformed into each individual’s structural brain MRI scan using a nonlinear template alignment, then coregistered to it. A probabilistic tag for the M1 cortical area (Talairach coordinates 35, − 28, 59), the stimulation site for MT determination and the corresponding skull position for the shortest distance from brain to skull were then displayed onto the native brain and used as landmarks for distance measurements. These tagged brain images were individually aligned with a virtual AC–PC line in an axial plane using MEDx (http://medx.sensor.com/products/medx) and the volumes resliced in 1 mm coronal sections. Using the method described previously [Kozel et al., 2000], one investigator (LF) performed all measurements twice on seven consecutive slices with the tag being in the middle. The shortest distances were acquired once in the direction of scalp-to-brain and the second in the direction of brain-to-scalp, both yielding an average distance skull to cortex (SCD).

Diffusion Tensor Imaging and Fiber-Tracking Analysis

DTI data was initially processed using FSL (http://www.fmrib.ox.ac.uk/fsl/). Eddy current distortion correction and diffusion tensor calculation were all performed using FSL [Smith et al., 2004]. A brain mask was calculated and a 3 × 3 voxel edge was eroded to minimize edge effects.

Fiber-tracking was performed using fiber-tracking software provided by the mrDiffusion package (http://sirl.stanford.edu/newlm/index.php/DTI). FSL-processed DTI images were resampled and interpolated to 2 mm isotropic voxels using a spline-based tensor interpolation algorithm. Fiber bundles only from the left hemisphere (hemisphere of interest) were estimated using a streamlines tracking algorithm [Dougherty et al., 2005]. Seed points were defined to identify the corticospinal tract in a modification of a standard method previously described and validated [Wakana et al., 2004]. As in the study by Wakana et al. [2004], fibers were tracked from seed points in a region manually defined around the posterior limb of the internal capsule (PIC) (identified at the level of the foramen of Monro) that intersect another region of interest manually drawn around the caudal portion of the cerebral peduncle (CP) (see Fig. 1).

Figure 1
Sample illustrations of ROI placement details. Panel A illustrates the posterior limb of the internal capsule (PIC) identified at the level of the foramen of Monro. Panel B shows ROI placement around the cerebral peduncle (CP). Finally, Panels C–E ...

Artifacts were eliminated by discarding fibers crossing midline. All ROIs were drawn by a single author (TH).

This fiber bundle was further constrained to increase anatomical validity [Axer and Keyserlingk, 2000] to include only a portion of this bundle lying within a centimeter inferior and superior to and 5 mm anterior to the center of mass of the PIC region of interest. This final fiber bundle was used to define a ROI (referred to as CST) corresponding to the corticospinal tract as it passes adjacent and lateral to the thalamus (see Fig. 2). It is important to emphasize that we elected to work with this somewhat distal segment of the corticospinal tract because of the increased reliability of fiber-tracking in this region.

Figure 2
The fiber bundle representing corticospinal tract, tracked with seeds at CP and PIC (white) and the constrained portion used to define CST ROI (blue) in two views. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com ...

However, because our primary interest lies with white matter adjacent to the motor cortex, we defined an additional ROI corresponding to the hand area of the motor cortex. A single author (TH) identified the hand knob on axial T2 images [Yousry et al., 1997]. The hand knob white-matter ROI was defined by selecting a seed voxel in the white matter directly adjacent to the middle of the hand knob gyrus, as identified on the T2-weighted axial images. The ROI (referred to as HK) was grown using an in-house 3-D flood-fill algorithm that selected all spatially contiguous voxels within a 1 cm radius of the seed voxel that also had an FA value within 15% of the FA of the seed voxel.

For each ROI described above, we calculated the mean fractional anisotropy (FA) [Taylor et al., 2004] and the mean principle diffusion direction (PDD) averaged across all voxels in a given ROI [Schwartzman et al., 2005]. The PDD of a voxel is the eigenvector corresponding to the largest eigenvalue of that voxel’s diffusion tensor. PDD is defined to be a unit vector and can be described in terms of three coordinates with respect to the major axes (see Fig. 3).

Figure 3
Orientation of the principle diffusion direction (PDD) as defined in this paper. Macroscopically, PDD is a unit vector uniquely defined by three coordinates (PDD-X, PDD-Y, PDD-Z) that define an mean trajectory of white matter fibers in a specific region ...

We did not coregister the DTI scans across subjects, because methods for doing this with fractional anisotropy are not well established. Instead, we computed a simple measure of intersubject head tilt by coregistering each subject’s B0 scan to a single arbitrarily chosen subject using FSL’s affine registration tool. A 4 × 4 affine 12 degree of freedom transformation matrix was derived for the transformation between each subject brain and the arbitrarily chosen reference subject. The three yaw (x, y and z) rotation angles derived per subject were then used to control for head tilt effects.

Statistical Analysis

The following techniques were used: descriptive statistics (including the mean, SD and median), Shapiro–Wilks test for normality, independent sample t-test and Pearson chi-square comparisons across schizophrenia patients and controls, stepwise linear regression repeated measure analysis of variance (ANOVA) with repeated contrasts, and graphical displays. All statistical tests were two-tailed using SPSS 16.0.

RESULTS

General

Descriptive statistics for the schizophrenia patients and matched control groups are presented in Table I for all demographic and clinical assessments. Groups showed significant differences in employment (schizophrenia subjects were less likely to be employed full time; P = 0.012), apathy evaluation scale (P = 0.06), and all MASQ scores.

TABLE I
Descriptive statistics for the schizophrenia patients and matched control groups for all demographic and clinical assessments

Table II compares several variables in the subject schizophrenia patient and control group. Subjects and controls did not differ in any significant way with respect to any of the measures related to cortical excitability, SCD or DTI-based measures. This homogeneity is the basis of our decision, in the analysis that follows, to pool subjects and controls and analyze them as a single group to increase power. The validity of the decision to pool subjects is further bolstered by the fact that clinical status is a nonsignificant regressor for MT in our model (see below).

TABLE II
Comparison of several variables in the subject schizophrenia patient and control group

rMT showed significant relationships with skull to cortex distance (SCD) (r = 0.83; P < 0.0001). MT was also significantly correlated to posterior internal capsule (PIC) principal diffusion direction Y axis component (PDD-Y) (r = −0.69; P = 0.002) and HK PDD-Y (r = −0.594; P = 0.012). MT was not significantly correlated with subjects’ age, PIC mean FA, PIC PDD-X, PIC PDD-Z, HK PDD-X or HK PDD-Z. The relationship between the proximal (HK) and distal (PIC) portions of the corticospinal tract was also compared. There was a significant correlation between PDD-Y of the PIC and HK ROIs (r = 0.583; P = 0.014). Finally, we calculated the correlations between the tilt angles and MT and the PDD-Y values. Tilt angles were not correlated to MT and PDD-Y values with one exception: TILT-Z and HK PDD-Y were correlated (r = −0.525; P = 0.03).

Intrarater Validity Analysis

The correlation between two sets of measurements for SCD done in opposite directions and performed by one rater (LF) was r = 0.76, P < 0.0001. The CST ROI intra-rater reliability was calculated by reproducing the ROI with the same rules for a subset of 11 subjects and correlations were strong (for PDD-Y: r = 0.9138, P < 0.0001; for FA: r = 0.8456, P = 0.0004). The corresponding PDD-Y and FA intrarater reliabilities for the HK ROI were not obtained.

Linear Regression

Following individual positive correlations with MT we explored the linear regression with MT as a dependent variable and clinical group (schizophrenia vs. control), SCD and PIC PDD-Y as the independent variables. The model (see Fig. 4) showed an overall highly significant relationship (df = 3, R2 = 0.824, F = 20.3, P < 0.001) between MT and SCD (P < 0.001) and PIC PDD-Y (P = 0.008). Clinical group was nonsignificant (P = 0.41). A linear regression using SCD, clinical group and HK PDD-Y as independent variables showed an overall highly significant relationship (df = 3, R2 = 0.767, F = 14.3, P < 0.001) between MT and SCD (P = 0.001) but not HK PPD-Y (P = 0.058). Clinical group was nonsignificant (P = 0.62).

Figure 4
Linear regression plot with MT as dependent variable and SCD and PIC PDD-Y as independent variable.

We also investigated each clinical subgroup with similar linear regressions (excepting clinical group as an independent variable). In the healthy control group, a linear regression with MT as the dependent variable and SCD, and PIC PDD-Y was significant (df = 2, R2 = 0.799, F = 11.8, P = 0.008) with SCD achieving significance as an independent variable (P = 0.012) but not PIC PDD-Y (P = 0.19). The same regression in the schizophrenia subgroup was also significant (df = 2, R2 = 0.894, F = 21.1, P = 0.004) with both SCD (P = 0.008) and PDD-Y (P = 0.008) achieving significance. Using HK PDD-Y instead of PIC PDD-Y in the subgroup regressions, we find that the regression is significant in the healthy control group (df = 2, R2 = 0.793, F = 11.5, P = 0.009) with only SCD achieving significance (P = 0.011) while not reaching significance in the schizophrenia subgroup (df = 2, R2 = 0.690, F = 5.57, P = 0.053).

DISCUSSION

To our knowledge, this study is the first to investigate the linear relationship between TMS MT and two anatomic variables in addition to age. While we again reconfirm the correlation between MT and SCD that is by now well established, we also report a novel DTI-based measure of white matter fiber orientation that is highly correlated with MT. Although some have empirically reported on the intimate relationship with MT and fiber orientation, this is the first in vivo, image-based report of the facilitation of MT with more forward-directed fibers. We found no significant relationship between MT and fractional anisotropy of the corticospinal tract. While this study uses data from a study of cortical excitability differences between unmedicated schizophrenics and controls, MT was not significantly different across groups and the schizophrenia status did not influence our results. We proposed a primary hypothesis with an intention to generalize our results and we combined data from both groups for added statistical power. With relatively small group samples, we did not find any statistically significant differences between our nonmedicated schizophrenia subjects and healthy controls, except for measures of apathy and employment status. This is likely due in part to a rigorous effort to match on gender, race, socio-economic status and education and previous history of substance abuse and is discussed in more detail in a separate publication.

It is generally agreed upon that TMS produces neuronal “excitation … where the negative-going first spatial derivative of the induced electric field parallel to the long axis of the nerve (i.e. the axon) peaks” [George and Belmaker, 2000]. Multiple early studies of TMS showed that magnetic field orientation was a critical factor in modulating neuronal excitation [Maccabee et al., 1993; Pascual-Leone et al., 1994a; Ruohonen et al., 1995, 1996]. Relative orientation of nerve fiber and the electric field are important factors in the excitation of shorter structures including interneurons, cell bodies and dendrites [Amassian et al., 1998]. As for longer nerve fibers such as pyramidal tract neurons, bends (see Fig. 3) and nonuniformities in the trajectory of the axon under the applied field seem to be the dominant factors in TMS-induced neuronal excitability [Abdeen and Stuchly, 1994; Amassian et al., 1992; Maccabee et al., 1993]. TMS achieves its effect largely by presynaptic, indirect (I-wave) activation of motor fibers [see George and Belmaker, 2000, p. 53]. In vivo functional TMS studies also show that orientation of magnetic field produces differential MEP and that a field orientation exists, which maximizes the MEP [Pascual-Leone et al., 1994b]. Chen et al. [2003] state that differential field directional effects in MEP generation are “likely related to different orientations of various groups of cortical fibers.” Different field directions recruit different action potentials. Chen et al. [2003] cite evidence that A–P (antero–posterior) currents activate cortical motor neurons indirectly leading to I-waves, and as such are different from L–M (latero–medial) currents, which lead to D-wave activation [Chen et al., 2003; Di Lazzaro et al., 1998; Kaneko et al., 1996]. These are not mutually exclusive; however, at stimulus intensities above the active MT, D-waves are also recruited by A–P-oriented current [Di Lazzaro et al., 1998].

Our results are consistent with these existing models of TMS neuronal effects and seem to suggest that close to 80% of the variance in MTs can be explained by SCD and white matter fiber orientation in cortico-pyramidal tracts. Previous studies have indicated that SCD alone accounts for ~50–60% of this variance [Kozel et al., 2000; McConnell et al., 2001; Stokes et al., 2005]. The measure for general orientation of the fiber tract introduced in this study, the mean principle diffusion direction of a ROI, is new to this literature. In our sample, SCD by itself accounted for 69% of the variance seen in MT (calculated by squaring the Pearson correlation coefficient). Similarly, the PIC PDD-Y explains ~48% of the variance in MT, and both PIC PDD-Y and HK PDD-Y are highly correlated. In the linear regression model, PIC PDD-Y accounts for an additional 13% of the variance in MT over and beyond that explained by SCD alone—a significant clinical result in analyses of this type. Schwartzman et al. [2005] first made use of PDD on a voxel-by-voxel basis to test for group differences. Our analysis differs in two significant ways: first, we calculate a mean PDD of all voxels in the ROI. Second, we use this statistic in a correlation. The physical interpretation of PDD, while relatively straight forward when applied to a specific voxel, is less clear in the context of a group mean. Conceptually, it gives a rough measure of the overall three dimensional orientation of a given bundle. Principal diffusion direction mean in an ROI is a macroscopic average, more likely a measure of long fiber orientation than short (inter) neuron or microscopic (dendrite, cell body) geometry.

In breaking down the three dimensional principal diffusion direction vector into orthogonal components, we found that only the A–P axis of the white fiber tracts is significantly correlated to MT, while the L–M axis is not. This is true at both the hand knob and the CST levels and is reflection of long track, and hence D-wave, electrophysiology. The low resolution of DTI scans does not allow us to make inferences about interneuron, hence I wave, electrophysiology. Such findings support previous research showing that D-waves are also generated by an A–P-induced current although such direction favors I waves. It is known that direct stimulation of corticospinal long tract fibers via D-wave activation relies primarily on electrical fields oriented coronally in the L–M plane. However, D-wave volleys are also observed when stimulation occurs in the A–P direction, although to a smaller extent, and more so at suprathreshold pulse intensities [Di Lazzaro et al., 1998]. That is why, when we restrict the induced electrical current to the A–P axis, as we did in this experiment, we limit the critical interplay to one of the three axes and in the same orientation. Indeed, because of the fixed coil position, it is not expected to find a significant relationship between MT and the two other planes, because they are orthogonal to the axis being stimulated. We conjecture that if a follow-up experiment was to be repeated with an L–M coil orientation, the PDD-X component would be a significant regressor, whereas the PDD-Y component would not.

DTI-Based White Matter Statistics and MT

In this paper, we propose a semiautomated, DTI-based ROI definition. Methodologically, DTI poses a unique set of issues in analysis and interpretation of data. DTI-based studies are still largely dependent on techniques employed in the analysis of more traditional MRI and CT structural information—mainly voxel-based morphometric techniques and ROI-based analysis. VBM-based analysis loses specificity but requires no a priori anatomical hypotheses. The central issue in the definition of an ROI is selection of appropriate region and manual area definition, which is simultaneously cumbersome and prone to issues of reliability. Here, we use a semiautomated technique requiring manual identification of two easily identifiable large-scale structures (the internal capsule and cerebral peduncle), each on only one axial slice. The remaining work, which involves cropping a fiber-tracked bundle, is entirely automated. This technique makes full use of the three-dimensional information encapsulated in a single diffusion tensor. By combining several different DTI methods, we are able to apply novel ROI definitions in a useful way.

Fiber-tracking algorithms are in their infancy and remain imperfect tools. Technical limitations of these algorithms led to our decision to analyze only a portion of the corticospinal tract around the internal capsule. Most importantly, the CST did not track to the entire precentral gyrus and did not consistently track to the hand area. This is a known limitation of currently used fiber-tracking algorithms stemming from the rich density of adjacent crossing white matter bundles. Traditional methods of defining a reliable ROI based on the hand region of the motor cortex were problematic for a few reasons. Such a method invariably requires a manual procedure to outline the ROI locate the hand knob or some other anatomical landmark. The hand knob is variable in morphology and not universally present [Yousry et al., 1997]. In addition, the boundaries of the relevant motor cortex are arbitrarily delimited with consequent decrease in specificity. Defining a proper ROI then becomes time-consuming and unreliable. This stands in contrast to identifying a portion of the internal capsule by fiber-tracking, a relatively simple and robust procedure. In our case, we are fortunate that the average fiber orientation in the internal capsule, given by the average principle diffusion direction, is highly correlated with average fiber orientation in the more anatomically relevant hand region of the motor cortex. Thus, we are able to use the more reliable ROI for our model. In addition, using tracked fiber bundles to define 3D ROIs offers the additional advantage of allowing ROI definition on a subject-by-subject basis. This method does not rely on a group template and therefore should be more sensitive to individual variations in morphology.

Contrary to previous reports, we did not find a significant relationship between age and MT. In comparison to previous studies, ours had a somewhat narrower range of ages and thus may have been underpowered to detect a smaller effect size.

Limitations

Our study has limitations. No attempt was made at registering our DTI scans. Our decision was made primarily because a valid methodology of registering images by fractional anisotropy or some other diffusion measure is not yet well established [Smith et al., 2006]. We have however provided evidence that the model is valid even when corrected for a measure of head tilt.

Our protocol uses multiple repeats of a six directional diffusion scanning protocol. At the time this study was designed this was the widely used standard. More recent research suggests that higher directional scans improve SNR and reproducibility of DTI scans [Landman et al., 2007]. Other simulations (Robert Dougherty, personal communications) indicate that scans with a lower number of directions and multiple repeats yield tensor estimates roughly as reliable as those with much higher numbers of directions. Nonetheless, future replication studies will take advantage of scans measuring more diffusion directions.

Our TMS coil placement method in the A–P axis did not use fixed plane markers. Because orientation is a significant regressor in our model, future studies of this kind should standardize coil orientation. In addition, although our center of mass for the HK region was anatomically located (see method above), the ROI was constrained to a 1 cm radius from the selected seed point to avoid partial volume effects near the cortical surface where diffuse white matter structures intermingle with gray matter. That decision may have introduced a larger variance than what is observed in CST and could account for the reduction in statistical significance observed in this region.

Although one rater was responsible of all SCD measurements [LF] and an average of two measurements was used for SCD, no formal intrarater validity analysis was attempted. The CST ROI, with its clear-cut axial slice selection rules and simple (and large) structural shapes involved (the internal capsule and cerebral peduncle), was robustly reproducible, whereas the HK ROI appeared comparatively less. Finally, no interrater reliability was calculated for any of the measurements in this study. This is a methodological concern in studies of this sort.

Future studies can clarify how cortico–spinal tract orientation influences MT by further studying the interplay of D-waves and coil orientation. A useful experiment would follow a similar design to ours but using a L–M coil orientation instead of the A–P orientation used here. Another possibility would involve fixing A–P coil orientation while varying suprathreshold field intensity.

SUMMARY

In this study, we present a linear regression model where two parameters, skull to cortex distance and a measure of motor bundle fiber orientation, account for more than 80% of the variance observed in TMS MT. TMS dosing is dependent on the coil position, the interplay with the underlying fiber tract orientation and the effective distance to reach the cortex. We also propose a novel application of DTI and fiber-tracking techniques to measure bundle fiber orientation. The validity of this method needs further evaluation but suggests that future applications of TMS over other various cortical regions may need to account for this important variable.

Acknowledgments

The authors would like to thank Drs. Scott Christie and Patricia Nnadi and Mr. Jeff Yungman for their assistance in recruiting subjects, Dr. Mark Eckert for his help in image processing and Mrs. Elaine Henry for her invaluable assistance and support.

Contract grant sponsor: NIMH; Contract grant number: MH065630-01A1.

Footnotes

Conflict of interest: Each of the authors submitted a detailed conflict of interest disclosure to the Journal.

References

  • Abdeen MA, Stuchly MA. Modeling of magnetic field stimulation of bent neurons. IEEE Trans Biomed Eng. 1994;41:1092–1095. [PubMed]
  • Amassian VE, Eberle L, Maccabee PJ, Cracco RQ. Modelling magnetic coil excitation of human cerebral cortex with a peripheral nerve immersed in a brain-shaped volume conductor: The significance of fiber bending in excitation. Electroencephalogr Clin Neurophysiol. 1992;85:291–301. [PubMed]
  • Amassian VE, Cracco RQ, Maccabee PJ. Basic Mechanisms of Magnetic Coil Excitation of Nervous System in Humans and Monkeys and Their Applications. New Orleans: McGregor and Warner; 1998. pp. 10–17.
  • Awiszus F. TMS and threshold hunting. Suppl Clin Neurophysiol. 2003;56:13–23. [PubMed]
  • Axer H, Keyserlingk DG. Mapping of fiber orientation in human internal capsule by means of polarized light and confocal scanning laser microscopy. J Neurosci Methods. 2000;94:165–175. [PubMed]
  • Basser PJ, Pierpaoli C. A simplified method to measure the diffusion tensor from seven MR images. Magn Reson Med. 1998;39:928–934. [PubMed]
  • Berg JW, Appelbaum PS, Grisso T. Constructing competence: Formulating standards of legal competence to make medical decisions. Rutgers Law Rev. 1996;48:345–371. [PubMed]
  • Chen R, Yung D, Li JY. Organization of ipsilateral excitatory and inhibitory pathways in the human motor cortex. J Neurophysiol. 2003;89:1256–1264. [PubMed]
  • Cicinelli P, Traversa R, Bassi A, Scivoletto G, Rossini PM. Interhemispheric differences of hand muscle representation in human motor cortex. Muscle Nerve. 1997;20:535–542. [PubMed]
  • Di Lazzaro V, Oliviero A, Profice P, Saturno E, Pilato F, Insola A, Mazzone P, Tonali P, Rothwell JC. Comparison of descending volleys evoked by transcranial magnetic and electric stimulation in conscious humans. Electroencephalogr Clin Neurophysiol. 1998;109:397–401. [PubMed]
  • Dougherty RF, Ben-Shachar M, Deutsch G, Potanina P, Bammer R, Wandell BA. Occipital–callosal pathways in children: Validation and atlas development. Ann N Y Acad Sci. 2005;1064:98–112. [PubMed]
  • George MS, Belmaker RH. Transcranial Stimulation in Neuropsychiatry. Washington, DC: American Psychiatric Press; 2000.
  • Harris-Love ML, Perez MA, Chen R, Cohen LG. Interhemispheric inhibition in distal and proximal arm representations in the primary motor cortex. J Neurophysiol. 2007;97:2511–2515. [PubMed]
  • Kaneko K, Kawai S, Fuchigami Y, Morita H, Ofuji A. The effect of current direction induced by transcranial magnetic stimulation on the corticospinal excitability in human brain. Electroencephalogr Clin Neurophysiol. 1996;101:478–482. [PubMed]
  • Kozel FA, Nahas Z, deBrux C, Molloy M, Lorberbaum JP, Bohning D, Risch SC, George MS. How coil–cortex distance relates to age, motor threshold, and antidepressant response to repetitive transcranial magnetic stimulation. J Neuropsychiatry Clin Neurosci. 2000;12:376–384. [PubMed]
  • Kubicki M, McCarley R, Westin CF, Park HJ, Maier S, Kikinis R, Jolesz FA, Shenton ME. A review of diffusion tensor imaging studies in schizophrenia. J Psychiatr Res. 2007;41(1/2):15–30. [PMC free article] [PubMed]
  • Landman BA, Farrell JA, Jones CK, Smith SA, Prince JL, Mori S. Effects of diffusion weighting schemes on the reproducibility of DTI-derived fractional anisotropy, mean diffusivity, and principal eigenvector measurements at 1.5 T. Neuroimage. 2007;36:1123–1138. [PubMed]
  • Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. MR imaging of intravoxel incoherent motions: Application to diffusion and perfusion in neurologic disorders. Radiology. 1986;161:401–407. [PubMed]
  • Maccabee PJ, Amassian VE, Eberle LP, Cracco RQ. Magnetic coil stimulation of straight and bent amphibian and mammalian peripheral nerve in vitro: Locus of excitation. J Physiol. 1993;460:201–219. [PubMed]
  • McConnell KA, Nahas Z, Shastri A, Lorberbaum JP, Kozel FA, Bohning DE, George MS. The transcranial magnetic stimulation motor threshold depends on the distance from coil to underlying cortex: A replication in healthy adults comparing two methods of assessing the distance to cortex. Biol Psychiatry. 2001;49:454–459. [PubMed]
  • Mills KR, Nithi KA. Corticomotor threshold to magnetic stimulation: Normal values and repeatability. Muscle Nerve. 1997;20:570–576. [PubMed]
  • Mishory A, Molnar C, Koola J, Li X, Kozel FA, Myrick H, Stroud Z, Nahas Z, George MS. The maximum-likelihood strategy for determining transcranial magnetic stimulation motor threshold, using parameter estimation by sequential testing is faster than conventional methods with similar precision. J ECT. 2004;20:160–165. [PubMed]
  • Pascual-Leone A, Cohen LG, Brasil-Neto JP, Hallett M. Non-invasive differentiation of motor cortical representation of hand muscles by mapping of optimal current directions. Electroencephalogr Clin Neurophysiol. 1994a;93:42–48. [PubMed]
  • Pascual-Leone A, Cohen LG, Brasil-Neto JP, Valls-Sole J, Hallett M. Differentiation of sensorimotor neuronal structures responsible for induction of motor evoked potentials, attenuation in detection of somatosensory stimuli, and induction of sensation of movement by mapping of optimal current directions. Electroencephalogr Clin Neurophysiol. 1994b;93:230–236. [PubMed]
  • Rossini PM, Barker AT, Berardelli A, Caramia MD, Caruso G, Cracco RQ, Dimitrijevic MR, Hallett M, Katayama Y, Lücking CH, Maertens de Noordhout AL, Marsden CD, Murray NMF, Rothwell JC, Swash M, Tomberg C. Non-invasive electrical and magnetic stimulation of the brain, spinal cord and roots: basic principles and procedures for routine clinical application. Report of an IFCN committee. Electroencephalogr Clin Neurophysiol. 1994;91:79–92. [PubMed]
  • Ruohonen J, Ilmoniemi RJ. Physical principles for TMS. In: Pascual-Leone A, Davey ND, Rothwell JC, Wassermann EM, Puri BK, editors. The Handbook of Magnetic Stimulation. London: Arnolds Publishers; 2002. pp. 18–29.
  • Ruohonen J, Ravazzani P, Grandori F. An analytical model to predict the electric field and excitation zones due to magnetic stimulation of peripheral nerves. IEEE Trans Biomed Eng. 1995;42:158–161. [PubMed]
  • Ruohonen J, Panizza M, Nilsson J, Ravazzani P, Grandori F, Tognola G. Transverse-field activation mechanism in magnetic stimulation of peripheral nerves. Electroencephalogr Clin Neurophysiol. 1996;101:167–174. [PubMed]
  • Schwartzman A, Dougherty RF, Taylor JE. Cross-subject comparison of principal diffusion direction maps. Magn Reson Med. 2005;53:1423–1431. [PubMed]
  • Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TE, Johansen-Berg H, Bannister PR, De Luca M, Brobnjak I, Flitney DE, Niazy RK, Saunders J, Vickers J, Zhang Y, De Stefano N, Brady JM, Mathews PM. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage. 2004;23(Suppl 1):S208–S219. [PubMed]
  • Smith SM, Jenkinson M, Johansen-Berg J, Rieclert D, Nichols TE, Mackay CE, Watkins KE, Ciccarelli O, Cader MZ, Mathews PM, Behrens TE. Tract-based spatial statistics: Voxelwise analysis of multi-subject diffusion data. Neuroimage. 2006;31:1487–1505. [PubMed]
  • Stokes MG, Chambers CD, Gould IC, Henderson TR, Janko NE, Allen NB, Mattingley JB. Simple metric for scaling motor threshold based on scalp–cortex distance: Application to studies using transcranial magnetic stimulation. J Neurophysiol. 2005;94:4520–4527. [PubMed]
  • Taylor WD, Hsu E, Krishnan KR, MacFall JR. Diffusion tensor imaging: Background, potential, and utility in psychiatric research. Biol Psychiatry. 2004;55:201–207. [PubMed]
  • Wakana S, Jiang H, Nagae-Poetscher LM, van Zijl PC, Mori S. Fiber tract-based atlas of human white matter anatomy. Radiology. 2004;230:77–87. [PubMed]
  • Yousry TA, Schmid UD, Alkadhi H, Schmidt D, Peraud A, Buettner A, Winkler P. Localization of the motor hand area to a knob on the precentral gyrus. A new landmark. Brain. 1997;120 (Pt 1):141–157. [PubMed]
  • Ziemann U, Lonnecker S, Steinhoff BJ, Paulus W. Effects of antiepileptic drugs on motor cortex excitability in humans: A transcranial magnetic stimulation study. Ann Neurol. 1996;40:367–378. [PubMed]