Hey all,
I’m looking through my resting-state data, and on a few subjects I see big slow-wave artifacts that seem related to non-focal eye movements. I haven’t had a lot of luck creating SSPs for them, and was wondering if anyone had some suggestions.
Incidentally, some of them have kind of noisy-looking PSDs, but it’s not consistent (and may be another issue - the run from the bottom PSD has less LF artifact and more HF contamination, the top PSD is from the raw file run also attached).
Hi Benjamin,
The artifacts you show here are really low frequency (<1Hz). You could probably use a high-pass filter to remove the worst of it. As for the broadband noise above 40Hz, I suspect muscle. You can drop me this recording on the scratch disk if you want me to have a look.
Beth
I can mess around and find a good filter frequency for the LF stuff, since I don’t care much about < 1 Hz stuff. I’m a bit more concerned about the muscle artifact on the resting-state data since I’d rather not just stick a low-pass filter on it like I did with my task data. I’ll drop that raw file on the disk if you’ll get a chance to take a look.
Hi Benjamin,
I looked at S03 and I was able to get a really nice SSP for the eye movements:
Artifacts menu -> SSP Generic -> Time window, All file; Event name, empty; Frequency band, [0.5,7]Hz
For the high freq stuff - it looks like this person has some dental work or is jaw clenching. Look at the topography of the PSD at a freq in the noise range:
There seems to be a corresponding MEG artifact and an SSP analysis reveals some bilateral, temporal stuff.
I can send you the script I used to find the artifact event and SSP. Please feel free to stop by the lab and look at this together. Do you see a similar artifact in this subject in other runs? If yes, it could be dental...
Beth
Hi Beth,
Thanks. I will try out the Generic SSP with those sort of parameters and see if it works for my other subjects with similar eye movements. I was using events with a time window rather than looking at lower frequency components in the whole file.
That subject does show the same bilateral temporal topography in the mid-40 Hz range for the 2nd and 3rd resting-state runs, but not the first one (though there is some stuff there, it’s not remotely as starkly apparent). I have some similar components that show up on some other subjects with HF noise issues.
If you could send me the script that would be great. I’ll email you about meeting up, which would be good.
FYI, there is now a “cookbook” of examples on how to use SSP removal of artifacts. Check it out at: http://neuroimage.usc.edu/brainstorm/Tutorials/SSPCookbook
Beth
