Empty ImageGridAmp and Time fields after source localization

Dear Brainstorm Forum,

I’m running into a problem when I try to create power maps following the tutorial https://neuroimage.usc.edu/brainstorm/Tutorials/RestingOmega#Artifact_cleaning. After preprocessing and source localizing the data (Compute sources [2016], Kernel only: one per file, dSPM) I keep coming up with empty power maps.

When I examine the source localized files in Matlab I can see that the Time and ImageGridAmp fields are empty (as if they were never converted to continuous files?). Here's a link to the data files for this subject: https://drive.google.com/file/d/1K3E9WnZ0Qk8Vu_f-WiBM9NkUWgvCaDq8/view?usp=sharing

As a sidenote, I’m using individual empty-room recordings to compute the Noise covariance for each subject and the head models were created from the individual subject’s anatomy using the options specified in the tutorial.

Any ideas would be greatly appreciated and thank you for your time!

I keep coming up with empty power maps.

Please describe what you mean with "empty power maps.
Do you mean: empty variables in the files, blank cortex surface when you open the files for display?
By "power maps" do you designate the minimum norm source maps, or the PSD values computed from these source maps?
Please post screen captures illustrating your request.

When I examine the source localized files in Matlab I can see that the Time and ImageGridAmp fields are empty (as if they were never converted to continuous files?)

This is normal for "kernel only":

Before reading the tutorial you are referring to, we recommend you start by following all the introduction tutorials (section "Get started") using the example dataset provided:
https://neuroimage.usc.edu/brainstorm/Tutorials

Before following

Thank you for your quick response. Yes, I was referring to the blank cortical surface when I open the PSD files computed from the source maps. At first I thought that this may have been related to the colormap thresholds but when I tried to adjust the thresholds I saw that the min and max values were 0.

Screen Shot 2021-12-01 at 8.20.14 AM

Thank you for your time and expertise.

I downloaded the package you sent, but I still miss some information in order to reproduce the error you reported:

  • The file S02post/tess_cortex_pial_low.mat (without it I can't open the source file)
  • The first MRI volume in the subject
  • The PSD file you can't open
  • The options that you used to compute the PSD file from the dSPM source file you sent (screen capture of the PSD process options)

Hi Francois,

Sorry about that. Here's a link to the files. Please let me know if anything is missing.
anat and data folders for Sub02post: https://drive.google.com/file/d/10r_PM_YkWnrWmNBuVY1rhS_JQxWSfW09/view?usp=sharing
Screenshots of PSD options: Screen Shot 2021-12-02 at 9.32.08 AM
Screen Shot 2021-12-02 at 9.32.23 AM

Thank you!

Hi Katie,
I reproduced the problem. Indeed I get this in the Matlab window when computing the PSD:

BST> Skipping window #1 because it contains a bad segment.
BST> Warning: Frequency band "delta" is empty.
BST> Warning: Frequency band "theta" is empty.
BST> Warning: Frequency band "alpha" is empty.
BST> Warning: Frequency band "beta" is empty.
BST> Warning: Frequency band "gamma1" is empty.
BST> Warning: Frequency band "gamma2" is empty.
BST> All values are null. Please check your input file.
BST> All values are null. Please check your input file.

Using the "physical" units for the PSD worked though:
MEG_3D_S02post_@rawsub-OticonS05_ses-20180816_task-AEF_run-09_meg_notch_low_PSD_19_4000ms_Power_FreqBands

So looks like the bug is with the "Normalized" units option in the PSD, which I introduced at the same time as the physical units. In this case, the frequencies are also normalized: going from 0 to 1. So the frequency bands would have to also be normalized accordingly and it's not currently the case...

So while this is a bug that we'll need to fix (possibly by removing that option altogether). I would recommend using Physical units for now.

P.S. I saw you used the 2016 version of compute sources, probably because it's older data that had already been processed, or comparing with previous results. If possible though, you may want to use the latest version.

@Marc.Lalancette
Thanks for looking into this issue.

I disabled this invalid combination of options:

@kmBecker
Update Brainstorm, and use only the recommended options in the entire pipeline ("Physical units" for the PSD, Compute sources [2018])

Awesome, thank you for your help Marc and Francois!