Participants

The local Ethics Committees of the University Medical Center of Freiburg, Germany approved the study (approval number 425/11), which was carried out according to the latest version of the Declaration of Helsinki. Twenty-four healthy volunteers aged 25.5±0.7 years (12 women, 12 men) participated after having given written informed consent. None of the participants had a history of neurological illnesses, took any CNS-active medication at the time of testing, or met one of the exclusion criteria published in the safety guidelines for TMS [28]. Twenty-two participants were right-handers and two participants were left-handers according to self-report. The study consisted of four experiments (Fig. 1) in partly separate groups of participants; experimental sessions were separated by at least one week to avoid possible carry-over effects.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Experimental design.

The study consisted of four experiments. The flexTMS stimulator allowed a comparison of the three conventional TMS waveforms applied by a single stimulation device (Expt. 1). We probed resting motor threshold (RMT), MEP amplitude, and stimulus-response-curves (SRC). In experiment 2, full-sine TMS pulses were compared with waveforms consisting of two concatenated half-sine pulses of identical current orientation. Expt. 3 examined how modifications of a given half-segment of the full-sine pulse impact on RMT. Changes in amplitude (Expt. 3a) and duration (Expt. 3b) of a given half-segment were tested separately. The duration of full-sine pulses was extended covering a longer range in experiment 4. See Methods for a detailed description of experimental procedures.

https://doi.org/10.1371/journal.pone.0115247.g001

Electromyographic recording

Participants were seated comfortably in an armchair with their stimulated hand resting on a cushion. MEPs were recorded by surface electromyography from the fully relaxed abductor pollicis brevis muscle of the non-dominant hand using silver/silver chloride electrodes (surface area, 263 mm²; AMBU, Ballerup, Denmark) and a bipolar belly-tendon montage. Data were band-pass filtered (20–2000 Hz), amplified (Ekida DC universal amplifier, EKIDA GmbH, Helmstadt, Germany), digitized at a sampling rate of 5 kHz (MICRO1401mkII data acquisition unit, Cambridge Electronic Design Ltd, Cambridge, UK), and stored on a personal computer for online visual display and later offline analysis using Signal Software version 3 (Cambridge Electronic Design Ltd, Cambridge, UK). Participants were asked to relax the target muscle throughout the experiment, and muscle relaxation was monitored using online visual feedback of electromyographic activity in the target muscle.

Transcranial magnetic stimulation

We performed TMS with a flexTMS stimulator (IMETUM, Garching, Germany). The technical details of this stimulation device have previously been described in detail [12]. The intrinsic resonance frequency of the flexTMS device results in an unaltered full-sine waveform of 160 µs duration, and half-sine pulses with 80 µs duration, correspondingly. Based on these standard waveforms we designed the pulses for this study using custom-made software [12]. Different types of coil current waveforms were created consisting of up-, down-, and hold-segments. The concatenation of these three segment types enables flexible design and application of pulse waveforms within certain technical limits [12]. Stimulus waveform templates generated by software were then transferred to the stimulation device.

We recorded coil current and electric field waveforms essentially as previously described [12]. Pulse shapes were applied with a figure-eight air core coil (P/N 510519, MAG & More GmbH, Munich, Germany). The coil current and electric field were measured with a Rogowski current probe (CWT60B, Power Electronics Measurements Ltd, Nottingham, UK) and a search coil, respectively. The search coil, made of a five-turn circular winding with outer diameter of 2 cm and inner diameter of 0.5 cm, was positioned parallel to the TMS coil plane with a distance of approximately one millimeter at the focus of the figure-eight coil. The search coil output voltage is proportional to the electric field [29]. Data were sampled at 200 MHz and digitally filtered to 5 MHz (−3 dB cutoff) using the finite impulse response filter tool in Igor Pro software (Wavemetrics, Lake Oswego, OR, USA). Coil currents and induced voltages of waveforms used in the experimental sessions are presented in Fig. 2. In the remaining figures, waveform pictograms are simplified for clarity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Waveforms applied by the flexTMS device.

Depicted are recorded waveforms used in the four experiments (see Methods). Traces show coil current waveform (black traces) and the induced voltage of the search coil (light blue traces). The search coil output voltage is proportional to the induced electric field. All pulse shapes were measured at 25% MSO. (A, B, C) Shown are recorded pulses for the half-sine, full-sine, and monophasic waveforms used in Expt. 1. These waveforms are commonly used in TMS. (D) Depicted is the two segment waveform consisting of two concatenated PA-oriented half-sines used in Expt. 2. (E) In Expt. 3a, the amplitude of a given half-segment of the full-sine pulse was modified. In the shown pulse waveform, the second half-segment was reduced to 50% of the unaltered AP-PA pulse. (F) In Expt. 3b, the duration of a given half-segment of the full-sine pulse was modified. In the illustrated pulse waveform, the second half-segment of the AP-PA pulse was shortened to 60 µs. (G) The symmetrically prolonged full-sine pulses probed in Expt. 4 induce an electric field waveform where the opposite phases are temporally separated.

https://doi.org/10.1371/journal.pone.0115247.g002

Transcranial magnetic stimulation

TMS targeted the non-dominant M1-HAND. Single pulses were applied trough a figure-eight shaped stimulation coil with an outer diameter of 100 mm (MAG & More GmbH, Munich, Germany) connected to the flexTMS device. TMS was triggered manually and administered to M1-HAND at inter-sweep intervals of 5–10 s. The stimulation coil was centered tangentially on the scalp over M1-HAND with its handle pointing in a posterior direction and laterally at an angle of approximately 45° away from the midline. At this coil orientation, TMS-induced current in M1-HAND was optimal for activating corticospinal neurons trans-synaptically via horizontal cortico-cortical connections [30]. Current direction was reversed, if necessary, from posterior-anterior (PA) to anterior-posterior (AP) by altering waveform polarity without changing coil position. Throughout the manuscript, current direction refers to induced current in M1-HAND. We located the optimal position for eliciting MEPs from the target muscle, defined as coil position over M1-HAND that produced reliable MEPs of >50 µV with lowest stimulation intensity. This coil position was recorded and maintained using a stereotactic, optically tracked navigation system as described [31].

Cortical motor threshold

The cortical resting motor threshold (RMT) was determined by a probabilistic threshold-hunting method using Motor Threshold Assessment Tool software (MTAT 2.0) [32], [33]. The TMS intensity to use for each trial is specified by the program and adjusted trial-by-trial based upon information whether an MEP was elicited or not. We always used 16 stimuli starting at 45% of maximum stimulator output (MSO) to determine RMT. The presence of an MEP was operationally defined as compound muscle action potential with peak-to-peak amplitude >50 µV. RMT was expressed as percentage of MSO. Within the four experiments, RMT measurements for the various stimulus conditions were performed in pseudo-randomized order.

In TMS, RMT can be used as measure of excitation threshold of the neuronal target elements [34]. For a given stimulation device, RMT thus directly reflects the effectiveness of neural stimulation, with lower RMT indicating a higher probability of successful suprathreshold membrane depolarization of neurons in M1-HAND [35]. The RMT is used in many studies to individually adjust the intensity of single-pulse and repetitive TMS, but the effectiveness of a given pulse waveform in rTMS protocols to induce lasting changes in cortical excitability might diverge from the effectiveness determined by RMT [21]. Since RMT is a reliable and physiologically relevant indicator of effectiveness of a single TMS pulse, we focused on measurements of RMT in the present study.

Experimental design

Data analysis

MEP peak-to-peak amplitude was determined with Signal software and averaged over 10 trials. Statistical analysis was done using SPSS version 18.0 Software (SPSS Inc., Chicago, IL, USA). Normality of data was confirmed using the Kolmogorov-Smirnov test. Statistical evaluation was performed via repeated-measures analysis of variance (rmANOVA). To compare the three principal waveforms (Expt. 1), rmANOVAs with the factor waveform (3 levels: monophasic, half-sine, and full-sine) were performed with RMT, MEP amplitude, and MEP latency as dependent variables. In Experiment 2, the within-subjects factor waveform (4 levels: AP-PA, PA-AP, PA-PA, and AP-AP) was entered for rmANOVA with RMT as dependent variable. For experiment 3, where the two segments of the full-sine waveform were varied separately, we used a three-way rmANOVA. For statistical analysis of this experiment, RMT data were normalized to the unaltered standard full-sine pulse. In Expt. 3a, pulse amplitude (3 levels: 75%, 50%, and 25%), current direction (2 levels: AP/PA and PA/AP) and pulse segment (2 levels: first and second half-segment) were used as within-subjects factors, and pulse length (2 levels: 140 µs and 180 µs), current direction (2 levels: AP/PA and PA/AP) and pulse segment (2 levels: first and second half-segment) in Experiment 3b. For Experiment 4, rmANOVA included the within-subjects factor duration (3 levels: 160 µs, 200 µs, and 240 µs). We used the Greenhouse-Geisser correction to adjust for violations of sphericity, if necessary. In case of significant interaction or main effects, rmANOVA was followed by separate two-way rmANOVAs, where applicable, and by post-hoc analysis using two-tailed Student’s paired or one-sample t tests. Multiple comparisons were corrected using the Bonferroni-Holm method. Level of statistical significance was set at P<0.05. Data are expressed as means ± sem except where stated.

Cortical excitability probed with different pulse waveforms

Different resonance frequencies of stimulators and possible differences in coil design may influence and partly obscure the specific effect of TMS waveforms. In our study, distinct pulse shapes were applied through the same TMS device [12], enabling a direct comparison of waveform effects without confounding factors due to differences in stimulator design. Measurements of RMT, MEP amplitudes, and stimulus-response-curves with half-sine, full-sine, as well as monophasic TMS waveforms showed consistently that full-sine pulses had the highest effectiveness to elicit an MEP (Fig. 3). RMT is used in this context as measure of a waveform’s effectiveness to elicit a motor response [27]. Greater effectiveness of full-sine stimulation has previously been reported comparing full-sine and monophasic TMS pulses [13], [20]–[24], [40]. Using the RMT as indicator of corticospinal pathway excitation, our results confirm and extend these previous studies by showing superior effectiveness of full-sine relative to half-sine and monophasic TMS waveforms using a single stimulator to apply all three waveforms. Importantly, we can thus rule out that differences in stimulator design have a pronounced contribution to the distinct effectiveness of monophasic, half-sine and full-sine stimuli. The impact of full-sine TMS on RMT prompted further investigations into how the configuration of the full-sine waveform determines its stimulation effect.

The shape of the full-sine waveform renders the TMS pulse more effective

When trying to assess which part of the full-sine waveform is responsible for stimulation effects or orientation-specific results, it has to be taken into account that the induced electric field E, which is determined by the rate of change of magnetic field over time [42], is most relevant for stimulation of neural tissue. Half-sine and full-sine TMS pulses induce an electric field that is more asymmetrical than that of the monophasic waveform (Fig. 2). This difference likely explains the stronger directional effects observed with monophasic TMS [13], [26]. When compared with the half-sine waveform, full-sine pulses are more effective and have similar directional effects with opposite initial current orientation. The full-sine pulse shape differs from half-sine in its duration and the induced electric field during the second half-segment of the waveform. The middle part of the full-sine waveform causes the longest change in magnetic flux having an opposite direction with respect to the first and third segment of the electric field waveform. This feature accounts for the direction-specific behavior of the unaltered full-sine pulse [13], [26].

Our findings provide direct experimental evidence that the shape of the waveform is responsible for the higher effectiveness of full-sine TMS pulses to elicit a motor response. Pulses with identical duration consisting of two concatenated half-sine segments had higher RMT than full-sine pulses and similar RMT to half-sine (Fig. 4). These results show that the shape of the full-sine waveform rather than its longer duration renders the full-sine waveform more effective than the half-sine waveform. The data also provide empirical evidence in support of the notion that the increased effectiveness in terms of M1-HAND stimulation is caused by the prolonged electric field phase in the middle part of the pulse [26], [27], [37], which is either positive or negative, depending on the initial current orientation of the full-sine TMS pulse.

Which part of the full-sine waveform contributes most to the stimulation?

The two half-segments with opposite current orientation cause the prolonged electric field middle phase of the full-sine waveform. Modulations of any given half-segment therefore result in alterations of this middle phase and may reveal the relative contribution of the two distinct half-segments. Elucidating which half-segment of the waveform contributes most to the stimulation effect is essential for our understanding of the physiology of full-sine TMS. To address this question, previous studies explored direction specific effects of TMS using monophasic and full-sine (biphasic) waveforms [13], [26]. Direction specific effects of full-sine pulses resembled those of PA monophasic pulses, if full-sine pulses had AP-PA orientation, and were similar to AP monophasic stimulation when full-sine TMS was PA-AP oriented [13]. From these results the authors concluded that the second half-segment of the full-sine waveform might be most relevant for direction-specific effects [13]. Up to now, technical limitations have hampered direct experimental testing of this hypothesis, which motivated the variation of full-sine pulse half-segments in the present study. For AP-PA and PA-AP current orientations, changing the pulse half-segment resulting in PA current flow in M1-HAND had the largest impact on RMT (Fig. 5). In single-pulse TMS, PA current direction has stronger stimulation effects compared with AP current direction for monophasic and half-sine waveforms, which most likely results from recruitment of different target structures with lower threshold [13]–[15], [24], [43], [44]. Our results show that the half-segment of the full-sine pulse inducing PA current in M1-HAND contributes more strongly to neural stimulation, regardless of whether it is the first or second segment. Interestingly, the amplitude reduction in the first and second half-segment had very similar effect on the middle phase of the electric field waveform. However, RMT was consistently changed to a greater extent if the half-segment inducing PA current was altered (Fig. 5). This is surprising, since amplitude and duration of the middle phase are likely to be critical factors determining the effectiveness of full-sine pulses. Thus, our findings provide experimental evidence that excitation of neuronal target cells probably occurs in the half-segment of the full-sine pulse inducing PA oriented current [44]. Another possible explanation for our results might be that the distinct phases of the full-sine waveform activate different populations of cortical neurons. Indeed, PA and AP oriented stimuli are assumed to target different neuronal structures with different relative thresholds [10], [19], [43]. Because full-sine stimuli are composed of half-segments with PA and AP current orientation, different neuronal targets are likely to play a role for TMS using full-sine pulses.

The changes of pulse amplitude in our experiments altered the middle part of the induced electric field waveform more than changes in pulse length did. This difference may explain why alterations of pulse segment duration influenced RMT less than changes of pulse amplitude did (Fig. 6). Another implication of our findings is that the amplitude of the half-sine pulse segment causing AP oriented current in M1-HAND can be markedly reduced with unchanged effectiveness of TMS. Amplitude reduction might be important considering that coil heating scales non-linearly by the power of two with coil current amplitude [45].

Implications for the stimulation effect of TMS

We also tested different pulses without altering the middle phase of the induced electric field waveform (Fig. 7). Although this part of the waveform remained unchanged, RMT decreased significantly with longer pulse duration. Thus, effectiveness of full-sine (biphasic) TMS is influenced by other parts of the waveform as well [13]. The additional electric field in the added waveform segments is nearly zero. Consequently, the negative and positive phases are separated by a short time interval. The two opposing segments of the full-sine pulse can be considered as primarily hyper- and depolarizing phases, and a sequence of hyperpolarization and subsequent depolarization was previously proposed as explanation for the higher effectiveness of full-sine stimulation [27]. The hyperpolarizing initial AP half-segment of a full-sine AP-PA pulse might lead to higher availability of sodium channels and thus enhance the depolarizing effect of the subsequent PA-oriented second half-segment [46]. Our results support this hypothesis by providing further experimental evidence. Alterations of the hyperpolarizing full-sine pulse half-segment inducing an AP oriented current had less effect than changes of the depolarizing PA half-segment (Fig. 5). Introducing a short delay between the negative and positive electric field phases reduced RMT (Fig. 7), which might act by enabling the hyperpolarization from the initial, negative electric field phase to subside more before the depolarizing (positive) middle phase. A recent modeling study demonstrated hyperpolarization effects by AP oriented parts of TMS waveforms for several different cortical neuron types [44]. The model could also reproduce the higher effectiveness of full-sine relative to monophasic pulses, and indicated that stimulation occurs through charge accumulation at axonal terminations [44]. Another possible explanation for the changes in RMT observed with the longer pulses in experiment 4 is that the hyper- and depolarizing segments of the waveform act as separated stimuli in short succession. Thus, the effectiveness of this pulse shape might also be due to temporal summation of excitatory postsynaptic potentials (EPSP) in M1-HAND.

The precise locus of excitation in M1-HAND and the specific neuronal target elements of TMS are still unresolved. Previous studies have addressed the locus of excitation using PET imaging [47] and modeling [48]. These data indicate that excitation and generation of early indirect waves (I-waves, [49]) likely occurs in cortical sulci [47] or their anterior wall [48]. Later I-waves might be generated at the crown of the precentral gyrus [48]. In our study, we compared the three commonly used TMS waveforms. The current orientation used for monophasic, half-sine, and full-sine pulses in our experiment 1 is thought to preferentially produce early I-waves [14], [39], probably originating from a similar site within M1-HAND [2], [48]. Accordingly, MEP latencies and stimulus-response curves for all three waveforms were comparable (Fig. 3E). The exact site of stimulation, however, is still not completely understood. And even within a certain cortical area, various types of neurons have different excitation thresholds for TMS [44]. It is thus conceivable that some of the observed threshold differences between waveforms are caused by recruitment of separate types of neurons in M1-HAND. In this context, our results elucidating the impact of different waveform parameters on the effectiveness to elicit a motor response may provide further important constraints for future modeling work.

Limitations of the study

Due to technical reasons, the flexTMS stimulator generated switching transients as well as oscillations in coil current after the end of the pulse, which were picked up as high-frequency voltage by the search coil (Fig. 2). The high frequency (cycle period ∼15 µs) of these oscillations makes a physiological contribution to the stimulation effect unlikely, because neuronal membranes act as a low-pass filter with much longer time constant (∼150–200 µs, [50], [51]). These oscillations were present for all waveforms and should not bias the observed differences between waveforms. We can, however, not fully exclude higher-order effects on the stimulation.

For our investigation of novel TMS waveforms we focused on the effectiveness of single pulses to excite human M1-HAND as quantified by RMT. Since we did not study any other measures of cortical excitability with the different waveforms used in the present study, such as active motor threshold or cortical silent period, we cannot exclude differential effects of the tested TMS waveforms on these measures. Furthermore, our study does not allow any direct conclusions on how effectively TMS can induce plasticity, which may be strongly influenced by waveform characteristics and current orientation of the TMS pulse [21].

Another important parameter for TMS is spatial focality. Focality is mainly determined by the shape and design of the stimulation coil [52], [53], but the pulse waveform may also influence the focality of TMS-induced cortex activation. Focality of TMS, i.e. the surface area activated by a single pulse, is of particular importance for functional cortical mapping defining its spatial resolution [54]. In this context, a flexible TMS pulse design allowing modifications of a given segment of the TMS pulse might be an interesting option to achieve higher focality. Since we did not investigate if any of the novel waveforms were associated with increased focality, this issue needs to be addressed in future studies.

Conclusion

In this study, we showed that the lower RMT of full-sine (biphasic) TMS is independent of stimulator design. Through flexible pulse design we were able to provide direct experimental evidence that the middle phase of the electric field waveform is critical for stimulation with full-sine pulses. Regarding the two oppositely directed half-segments of the full-sine waveform, we found that the relative contribution depends on current orientation of the pulse. The half-segment inducing a PA current is most important for the stimulation effect of full-sine pulses. Furthermore, RMT decreases with increasing duration of the full-sine waveform. Our findings contribute to a better understanding of the physiological effects of the biphasic full-sine waveform and might help improve future stimulator design for more efficient and precise non-invasive stimulation of the human brain.