Non-zero residual gradient area following biphasic TMS pulses

[BenInglis]

We have determined that a biphasic TMS pulse produced by the MagPro box is not perfectly balanced such that the positive and negative gradient lobes are not exactly of equal area but opposite sign. The result is a residual dephasing effect imparted to any magnetization that exists when the TMS pulse occurs.

[BenInglis]

Tests on 10/14/2015

SETUP: Birdcage coil, FBIRN gel phantom with TMS coil placed on top, slightly towards the front of the sphere. Biphasic TMS pulse of duration 250 usec, set at 70% amplitude, one TMS pulse per TR incremented 0.1 ms each time. Timing of TMS pulses and EPI sequence as listed in Issue #5.

DATA FILES: tmsMRI_sequence_70percentTMS_1for1TR, scan 7. tmsMRI_sequence_70percentTMS_1for1TR, scan 8. tmsMRI_sequence_70percentTMS_1for1TR_Sag, scan 9.

RESULTS: The TMS pulses clearly trace out the timing of the EPI pulse sequence. For the first approx. 35 TRs the TMS pulse hits the latter part of k-space from the 6th slice, then it goes through the first crusher gradient of the 7th slice. (This is the target zone for single pulse TMS with fMRI.) The fatsat pulse in the 7th slice shows up as a clear Gaussian modulation of signal in an ROI placed in any slice. Mean curve analysis on the scanner, using a circular ROI in slice 7 shows that during and at all times after the excitation RF pulse there is a step down in SNR (yellow trace):

The step down in SNR corresponds to a missing chunk of signal in the phantom section adjacent to the location of the TMS coil. The initial diagnosis was that a TMS pulse applied after creation of magnetization for imaging has a residual dephasing effect, i.e. that the biphasic TMS pulse is not properly balanced. The diagnosis was supported by repeating the measurements but with the TMS amplitude reduced to 50%. The step down in SNR was reduced, commensurate with a reduced net gradient imparted by the TMS pulses:

Finally, the image axis was changed to sagittal, with the TMS intensity left at 50%. The signal drop increased compared to the previous axial slices, indicating a different component of the TMS residual field is interacting with the imaging gradients:

CONCLUSION: A bipasic TMS pulse of duration 250 usec is insufficiently balanced to be applied after the creation of sample magnetization for imaging. It will not be possible to pulse the TMS after RF excitation without corrupting the current imaging slice. There will always be a region of dropout, corresponding to the remnant field of the unbalanced TMS, whether the TMS is pulsed during k-space readout or before. For the time being we will have to restrict TMS pulses to being coincident with the crusher gradients either side of the fatsat pulse.

[BenInglis]

Tests on 10/15/2015

SETUP: 12ch coil, FBIRN gel phantom with TMS coil placed on top, slightly towards the front of the sphere. Biphasic TMS pulse of duration 250 usec, set at 70% amplitude, one TMS pulse per TR incremented 0.1 ms each time. Timing of TMS pulses and EPI sequence as listed in Issue #5.

DATA FILE: tmsMRI_sequence_70percentTMS_1for1TR, scan 3.

RESULTS: Same results as for the birdcage coil in the previous comment.

CONCLUSION: The diagnosis given in the prior comment stands. There are small effects of RF coil selection but the step down in SNR for TMS pulses after creation of magnetization is due to a residual magnetic field after biphasic pulse. It doesn't appear that a conductive surface, e.g. for the creation of eddy currents, is required to observe the effect. The step down in SNR and the corresponding region of dropout is roughly constant regardless of the relative timing of the TMS pulse to the RF excitation.

[BenInglis]

Tests on 10/15/2015

SETUP: 32ch coil, FBIRN gel phantom with TMS coil forced into place atop the phantom. Biphasic TMS pulse of duration 250 usec, set at 70% amplitude, one TMS pulse per TR incremented 0.1 ms each time. Timing of TMS pulses and EPI sequence as listed in Issue #5.

DATA FILE: tmsMRI_sequence_70percentTMS_1for1TR, scan 2.

RESULTS: Same results as for the 12ch coil in the previous comment.

CONCLUSION: The diagnosis given in the prior two comments stands. There are small effects of RF coil selection but the step down in SNR for TMS pulses after creation of magnetization is due to a residual magnetic field after biphasic pulse. It doesn't appear that a conductive surface, e.g. for the creation of eddy currents, is required to observe the effect. The step down in SNR and the corresponding region of dropout is roughly constant regardless of the relative timing of the TMS pulse to the RF excitation. The region of dropout is located under the location of the TMS pulse, supporting the assignment of a residual non-zero gradient area from the TMS pulse rather than an eddy current in either the RF coil or the magnet cryostat.