search

sccn / roiconnect

ROI connectivity analysis in EEG
40 stars 17 forks source link

Unrecognized field name "pnts" #74

Open amisepa opened 8 months ago

amisepa commented 8 months ago

I successfully computed power at brain area level, but get this error for connectivity. I tried CS, aCOH, MIM. Same for snippets mode. 128-channel EEG data, sampling rate of 128 Hz, segmented 2-s epochs. EEG.roi.pnts does not exist for some reason. Just did a pull to make sure I had latest version.

Error in roi_connect (line 151) conn_mult = data2spwctrgc(source_roi_data, EEG.roi.pnts, g.morder, 0, g.naccu, [], tmpMethods2, [], 'freqresolution', g.freqresolution);

Error in pop_roi_connect (line 344) EEG = roi_connect(EEG, 'morder', g.morder, 'naccu', g.naccu, 'methods', g.methods,'freqresolution', g.freqresolution, 'roi_selection', g.roi_selection);

nguyen-td commented 8 months ago

Hi, thanks for your report. I just tested the connectivity estimation with CS, aCOH and MIM myself and I did not run into any issues. To help us reproduce the error, can I ask you for two things:

  1. We have some test pipelines in test_pipes/, the simplest one should be pipeline_connectivity. Could you please run it and tell me if the pop_roi_connect(...) part runs through? You may need to adjust the measures variable.
  2. EEG.roi.pnts should not be deleted at any point after computing the source time series through pop_roi_activity. Would you mind backtracking at which point this field is deleted?
amisepa commented 8 months ago

It worked, I was using the wrong code I had modified to run pop_roi_activity on snippets, some months ago because it wasn't work properly.

Is it necessary to always downsample to 100 Hz for ROIconnect?

nguyen-td commented 8 months ago

The last question is something for @stefanhaufe or @arnodelorme to answer.