Artifacted epoch rejection tools and visualization

Discussions
1

Dear BST Masters,

It happens to me that even after accurate rejection of bad segments by visual inspection of raw data, and EOG/ECG artifact removal with SSP, there are still a few epochs containing residual artifacts that I would like to reject.

To avoid another round of visual inspection, or at least be guided by some objective measures, I have tried to use the peak-to-peak threshold detection tool, but results are really poor on my Elekta MEG data (for some values, good epochs may be rejected while bad epochs may not).

So I wondered:

  1. If you have or are developing more sophisticated tools than peak-to-peak threshold detection, for example based on the probability distribution of single channel (or all channel) values.

  2. If not and I would like to implement one, what’s the simplest way to integrate it in BST - for example, is there a BST function I can start from to avoid writing it from scratch?

  3. About epoch visual inspection. Since I tend to visualize always the same features (time course and topography separately for magnetometers and gradiometers), and each time adapt the windows to my personal screen and preferred view, I wonder whether there is a way to implement this personal visualization with a click or a matlab command. This would be much appreciated also for continuous raw data reviewing.

  4. Again about epoch visual inspection. To scroll epochs from one condition to the next, I have to close and re-open all the windows (time courses and topographies) (F3 does not allow me to scroll between conditions), which in my study is really annoying because I have many different conditions. Is there a way to overcome this?

Thanks a lot for your feedback,

Best,

Marco

2

Hi Marco,

1-2) We implement new detection algorithms when we need them in our data analysis at the McGill MEG lab. At the present time, the automatic detection of ECG/EOG artifacts works well for us.
You are welcome to improve those functions, or the peak-to-peak threshold process, and share with us your code. The corresponding functions are:

  • process_evt_detect_ecg, process_evt_detect_eog, process_evt_detect (run on continuous files)
  • process_detectbad (runs on epoched data)

Make a copy of the process function in your personal process folder: $HOME/.brainstorm/process (this folder is always on top of your path).
Rename the file and the process, make the necessary modifications, send me the file, and I can add it to the Brainstorm distribution.

  1. It’s a feature requested by lots of users. We will add a sort of custom desktop feature where you can open and position all the figures you need, save the configuration, and be able to load it back with your next file. It will be in the new “Layout” menu in the right part of the toolbar. Development has not started yet, hopefully I’ll have to do it by the end of the year.

  2. Right-click on the figure > Navigator > Next/previous condition.
    Shortcut F2 / Sihft+F2

Cheers,
Francois

3

Ok thanks!

Marco

[QUOTE=Francois;5262]Hi Marco,

1-2) We implement new detection algorithms when we need them in our data analysis at the McGill MEG lab. At the present time, the automatic detection of ECG/EOG artifacts works well for us.
You are welcome to improve those functions, or the peak-to-peak threshold process, and share with us your code. The corresponding functions are:

  • process_evt_detect_ecg, process_evt_detect_eog, process_evt_detect (run on continuous files)
  • process_detectbad (runs on epoched data)

Make a copy of the process function in your personal process folder: $HOME/.brainstorm/process (this folder is always on top of your path).
Rename the file and the process, make the necessary modifications, send me the file, and I can add it to the Brainstorm distribution.

  1. It’s a feature requested by lots of users. We will add a sort of custom desktop feature where you can open and position all the figures you need, save the configuration, and be able to load it back with your next file. It will be in the new “Layout” menu in the right part of the toolbar. Development has not started yet, hopefully I’ll have to do it by the end of the year.

  2. Right-click on the figure > Navigator > Next/previous condition.
    Shortcut F2 / Sihft+F2

Cheers,
Francois[/QUOTE]