Inverse Solution Issue - mne 2009

Bugs
1

I am having issues with the 2009 mne solution. The solution does not correspond with the changing topography maps and does not look to reflect the underlying physiology. When performing a scout timeseries, all of the traces have the same shape and are scaled copies. The distributed map also appears to be the same across epochs. It also does not matter if I use the option of a kernel or full results. This appears to be an issue that happens with some studies, but not all. There are no errors during the processing. Also notable is that there is no problem when using the 2016 or 2018 mne solution.

Matlab=R2018a
Brainstorm=April 2019 or June 2019 (both have the same issue)
Scanner= Elekta

Slide2
Slide21280×720 427 KB
Slide3
Slide31280×720 422 KB
Slide1
Slide11280×720 271 KB

Let me know if you need any more information to help diagnose.

Thank you,
Jeff Stout

2

I am not sure if this is helpful, but here is a picture of the imaging kernels.

Imagekernel
Imagekernel1280×720 574 KB

Thanks in advance for any help on this issue.
-Jeff

3

I have been able to solve this issue. It does not look like the inverse solution is the problem. The emptyroom noise covariance matrix that was calculated was corrupted. This was an operation of the SSS process (even though I couldn't find any bad channels). I believe that the emptyroom noise dominated the inverse solution and caused the issue that I am seeing.

NoiseCovar
NoiseCovar1280×720 607 KB

Sorry for flagging this as a brainstorm issue.

Thanks,
Jeff Stout

4

The process "Detect cHPI activity" can help you detect and mark as bad these SSS transitions:
https://neuroimage.usc.edu/brainstorm/Tutorials/VisualSingle#Pre-processing

5

Thanks Francois.