Source time series/Parcellation time series

Discussions
, , ,
1

Hello,
I am a newbie to Brainstorm and source localisation. I have a long continuous EEG recording (Epileptic), nearly an hour. I wanted to source localise using this data. I built mne-sLORETA model. I am encountering few problems. Mu ultimate aim is to get times series for parcellated regions (68 regions that I parcellated cortex using freesurfer).

  1. I would like to have source time series for 15000 sources. I searched the forum and found one relevant discussion. However, I am unable to reproduce what Francosis has suggested there as in "Output Mode" only "Inverse Kernel" gets highlighted ("Full Results" cannot be selected). May I know why is this happening?

    BS_1
    BS_1894×510 113 KB

  2. I also tried the second procedure (in_bst_results) suggested by Francosis, even this one did not work. Unfortunately, I couldn't see the EEG time series at all when I imported into MATLAB (also checked in database folder, the filtered file is only 7KB). Can you please help me with this?

  3. My ultimate aim is to get time series for 68 parcellated (parcellated) FreeSurfer. I down-sampled 15002 vertices to 68 regions-Desikan-Killiany (Scouts/ROIs). Even here, I tried to export/save scout time series but couldnt find the matrix (68xtime points) either in MATLAB workspace or on hard-drive.

In case these questions were answered earlier, please accept my apologies and point me to those. Any help is much appreciated.
Regards,
Sri.

2

You probably right-clicked on the channel file or the head model. In this case, the source model created is necessarily shared across multiple recordings files, hence the "Inverse kernel" option enforced.
If you want the full sources time series, you need to right-click directly on the recordings for which you want to estimate the sources.
Watch out: if you do this on the full hours of recordings, this would try to create a gigantic (and probably useless) file. At 500Hz, you get something like 27 channels * 3600s * 500Hz * 15000 sources * 8 bytes > 5400 Gb !

I also tried the second procedure (in_bst_results) suggested by Francosis, even this one did not work.

What did you try?

Unfortunately, I couldn't see the EEG time series at all when I imported into MATLAB (also checked in database folder, the filtered file is only 7KB). Can you please help me with this?

You probably look at a link to a continuous file. It contains only some metadata explaining what to do.

My ultimate aim is to get time series for 68 parcellated (parcellated) FreeSurfer. I down-sampled 15002 vertices to 68 regions-Desikan-Killiany (Scouts/ROIs). Even here, I tried to export/save scout time series but couldnt find the matrix (68xtime points) either in MATLAB workspace or on hard-drive.

You won't be able to process the full hours of recordings at once, this represents volumes of data that are not manageable, even on computation cluster. Processing ongoing recordings at the source level is a complicated question. There are some measures that we optimized (read this tutorial: https://neuroimage.usc.edu/brainstorm/Tutorials/RestingOmega), but the extraction of scouts time series has not been optimized in this way yet.

If you really need the full hour for the 68 scouts, you can maybe split the recordings in smaller blocks (use the split option when importing the recordings), downsample the sLORETA source maps for each epoch, then concatenate again the reduced sources.

If you get issues with any step, please post details about these issues: describe what you try to get, the exact sequence of operations you do (with screen captures) and the results you get (with a comment on why it does not match your expectation).

Cheers
Francois

3

Hello Francois,
Many thanks for the reply. I managed to achieve the what I asked for (68 regions x timepoints). I followed the tutorial properly and found this process_extract_scout : Process "Extract > Scouts time series", which solved my problem.
I have another problem of EEG electrode co-ordinates. I am unable to find coordiantes for F11 F12 P11 P12 T3 T4 T5 T6.I have been searching for them for a quiet a while now, but couldn't get the locations of it. I tried contacting the company, unfortunately nothing has happened.
Searching on Brainstorm forum, I found a post in which you suggested to do it manually, however, this post is from 2010. I am just wondering if you have them now, that could really help me.
Regards,
Sri.

4

No, I'm sorry, these electrodes are not included in any templates we have: they are not part of the standard 10-10 (or 10-05) positions, and not included in any of the electrode caps for which we provide default positions.
You will have to place them manully on the scalp. Create an empty folder, set one of the existing templates as the channel file, right-click on the channel file > MRI registration > Edit. Then add the missing electrodes, save the modifications, copy this channel file somewhere else, and then load its positions in your recordings with the menu "Add EEG positions".

If you create a standard channel file including them (maybe the most complete Generic>10-05 one), please share it with us, I could include it in the Brainstorm distribution.

Cheers
Francois

5

Thanks for the reply, Francois. Will update you once I am finished on those.
Regards,
Sri.