Transcranial magnetic stimulation (TMS) uses transient and strong magnetic fields to elicit neuronal activity non-invasively ¹. Combining TMS with fMRI provides the unique opportunity to not only observe but also manipulate brain responses. Different from TMS-EEG, TMS-fMRI holds the promise of detecting deep brain responses to exogenous magnetic field stimulation.

Here we report our recent advance in TMS-fMRI technical development. Specifically, we tested how TMS pulses and the separation between TMS pulse and RF excitation affect MRI SNR and time-domain SNR (tSNR) using our tailored 8-channel TMS-compatible MRI coil array ². Results from an in vivo experiment using this integrated system to record the BOLD signal elicited by TMS pulses at human primary motor cortex is also reported. Taken together, our TMS-compatible MRI coil array can be a valuable tool in studying the immediate hemodynamic responses to TMS with high sensitivity.

An 8-channel surface coil array was designed and tested on a 3T MR system (Skyra, Siemens). All loop coils (center coil diamter = 9 cm, side coils diameter = 7 cm) were tuned to 123.25 MHz and connected to a low noise pre-amplifier (LNA) integrated with a mixer through a coaxial cable and a matching circuit, which transformed the impedance to 50 Ω in order to obtain the lowest noise figure. An active detuning circuit was formed using a variable inductor and a PIN diode. To minimize the coil array thickness for the maximal TMS stimulation efficacy, the coil inductor loops was formed by copper foil tape. The matching circuit and LNA were extended outside the coil loops through coaxial cables with 10-20 cm in length. Coil loops were hexagonally arranged on the mechanical housing (Fortus, 400mc) fitting to the TMS coil (MRI-B91, MagVenture). A hole in the center of coil array was designed to minimize the distance between the tangential plane of the head and the TMS coil to less than one millimeter. To mutually decouple between neighboring coils in the array, coils were critically overlapped. We first tested how MRI quality was affected by the presence of the TMS coil. Subsequently, we evaluated how TMS pulses modulated MRI quality. Both experiments used an EPI pulse sequence (FoV: 224x224 mm, Resolution: 3.5 mm isotropic, TR: 2000 ms, TE: 30 ms, Flip angle: 90^o). Two delays were added between consecutive EPI acquisitions. Type 1 had a delay of 1 s, including 0.7 s TMS pulses and 0.3 s silence; Type 2 had a delay of 0.75 s, including 0.7 s TMS pulses and 0.05 s silence (Figure 1B). One subject participated the in vivo experiment. We compared the SNR and tSNR maps using data acquired from our coil array and a commercial 32-channel whole-head array. TMS stimulation (110% of the subject’s motor threshold) at the human left M1 was delivered to the subject inside MRI using a blocked design (5 “on” and 5 “off” 30 s blocks). BOLD activities were estimated using the General Linear Model.Figure 1A shows SNR and tSNR maps using a commercial 32-channel whole-head array and our 8-channel TMS-compatible array. Both showed higher sensitivity at the bottom of the phantom. Note that the phantom sat directly on top of our 8-channel array. Figure 1C shows the SNR and tSNR profiles acquired from the middle line (white dash line) of all images. The improved SNR and tSNR by our array was found significant at depth no farther than 4 cm. With the TMS coil combination, both SNR and tSNR maps changed minimally. There was no visible distinction in SNR and tSNR maps between delivering TMS pulses (both type 1 and type 2) and delivering no TMS pulse. Figure 2 shows the experimental setup and the SNR as well as tSNR maps at an axial slice passing through the primary sensorimotor cortex. We found significant BOLD signal at the left hemisphere primary sensorimotor area, whose hemodynamic response shows good agreement with the expected fluctuation due to blocks of TMS pulse delivery (Figure 3).Our preliminary results successfully demonstrate that our TMS-compatible MRI coil caused neither visible image distortion nor SNR and tSNR degradation. The temporal separation between TMS pulses and the onset of EPI acquisition can be as short as 50 ms without clear effects. This suggests that the delivery of TMS pulses can be tightly packed between silent intervals of MRI acquisition. Further experiments at other functional areas, such as visual cortex, will be attempted to investigate the brain activity due to phosphene induced by TMS.