Understanding Transcranial Magnetic Stimulation: A New Study of High-Temporal-Resolution Cortical Single-Neuron Responses With Extensive Artifact Reduction

Transcranial magnetic stimulation (TMS) is a widely used method for stimulating neocortical regions in patients with medication-resistant depression and in research studies of other diseases. However, this technique is not standardized across centers and not well understood in terms of the relationship between mechanisms and effects. A few prior attempts have been made to relate TMS to underlying neural activity. Previous studies reported that cerebral blood flow increases in the presumed regions under TMS activation. Other groups have directly observed single-neuron activity during TMS, but the neuron signal recordings were hampered by strong TMS artifacts. Mueller et al¹ have now developed a new type of TMS coil, a modified recording system, and several analysis methods to suppress the influence of TMS artifact in neuronal recordings. After applying these techniques to minimize the artifact, the authors performed direct recordings of single-unit activity and studied the effects of different stimulation intensities.

The design processofthe coil included the use of finite-element models to concentrate the electric field strength and focality in a relatively isolated volume. Compared with a standard Magstim butterfly coil, the designed coil had comparable induced electric fields with slightly lower field magnitude and longer stimulation duration. The mitigation strategies for reducing TMS-induced signal artifacts included voltage artifact reduction techniques, the suppression of induced currents, and the management of coil vibrations. Voltage artifacts were reduced mainly by clipping the artifact with diodes before amplification and a software artifact subtraction approach (which detected the artifact waveform in an external coil and subtracted it from the recorded signal). The induced current artifact was evaluated through theoretical calculations and empirical measurements, both in vitro and in vivo, to confirm that it was well below the neuronal activation threshold. Coil vibration was dampened by adding insulation between the connection joints from the coil to the primate chair. Additionally, an offline template subtraction technique was implemented to further remove vibration artifact. These procedures allowed the total artifact duration related to the TMS pulse to be confined within 0.7 milliseconds.

Figure.

A, raw data from 20 sequential transcranial magnetic stimulation (TMS) pulses (70% active, top) and the sham control (bottom) at the same location in 1 animal. Labeling of the hypothesized neural substrates is provided. B, Raster plots of recorded spike times, active and sham stimulation, with an example spike waveform shown in the inset. C through E, further individual neuronal response to applied TMS. C, this neuron shows a quick short latency activation, followed by inhibition and then a burst. D, this example neuron shows no immediate response but delayed latency activity. E, this example neuron shows suppression of activity (gray bar) after the TMS pulse. Reprinted by permission from Macmillan Publishers Ltd: [Nat Neurosci] (Simultaneous transcranial magnetic stimulation and single-neuron recording in alert non-human primates. Nat Neurosci. 2014 Aug;17 [8]:1130–1136.), copyright (2014).

The single-unit activity was recorded from 2 rhesus macaques with implantation of nonfer-rous recording chambers. Both active and sham TMS consisted of 12 trials with 12-second interstimulus intervals. Simultaneous magnetic stimulation and neuronal recording showed higher increases of spike firing rates after active TMS compared with sham TMS (Figure). The neural population (frontal eye field) response to active TMS lasted approximately 100 milliseconds or longer. High-intensity active TMS elicited higher spike firing rates than low-intensity active TMS stimulation, low-intensity sham TMS stimulation, and high-intensity sham TMS stimulation. There were no significant differences between responses to high and low intensity for sham TMS or to active and sham TMS at low intensity.

Using these described procedures to mitigate artifacts induced by TMS, Mueller et al were able to observe clear and direct TMS effects on singleunit neural activity. Among the artifacts, the most worrisome one is the induced current because it may be mistakenly recognized as neuronal activity. After careful monitoring and mitigation, the induced current was reduced to approximately 80 nA (well below the magnitude that could result in direct neuronal activation). These methods could shed light on the basic effect of TMS on the neuron level and possibly guide and optimize future clinical applications of TMS.