[Eeglablist] TMS artifact removal

[Jeffrey Johnson]

Greetings-

I am currently running a study using combined TMS/EEG to assess the
event-related spectral perturbation resulting from a single TMS pulse, and
am looking for some assistance in dealing with the rather large artifact
produced by the TMS pulse.

We are using a Magstim Rapid II stimulating coil and a BrainAmp MRPlus
amplifier with a 64-Channel Ag/AgCl electrode cap. We are using a
'continuous recording' method, as opposed to a 'sample and hold' method
where input to the amplifier is blocked very briefly before and after the
TMS pulse is delivered. With the continuous recording method, the TMS pulse
creates a high amplitude artifact that lasts for ~10-15ms. However, when the
ERSP is computed, the TMS artifact gets carried well into the post-TMS time
interval, presumably as a result of the windowing procedure that is used to
determine spectral power. Thus, we are unable to obtain artifact-free ERSPs
from the time interval around ~20-80ms post-TMS, which is what we are
primarily interested in. The artifact can be attenuated to some extent using
ICA, but not eliminated, since it appears in most ICA components and is
found over the majority of the scalp. Also, the artifact does not appear to
be identical on every trial, or at each electrode, making it difficult to
simply subtract it from the data.

Our intuition tells us that it should be possible to effectively ignore the
time period containing the artifact in our calculation of ERSP. Thus far, we
have attempted zero-padding the interval from -6ms to +12 ms around the TMS
pulse, but this wasn't very satisfactory. Is there any satisfactory method
for interpolating values from one side of the artifact to the other, or
perhaps a more sensible method of dealing with this?

Any help would be greatly appreciated.

Best,
Jeff