Find_eog_events thresh value

đź’¬ Support & Discussions
1

Hi all,
I’m using find_eog_events and it’s finding way too many eye events for the MEG (neuromag, sometimes grad, sometimes mag) channels I’m using. Anyone have a suggestion for a thresh value I could try instead?
On a sample dataset, I tried a few values of thresh but every time it said no significant peaks found. Using no thresh setting, it finds 176 peaks.

I looked at an old post thresh units question where there was mention of advice to manually set the thresh option. He says he was using 90 for EEG data. I’ve looked a bit at the referenced peak_finder module, so I know without a value for thresh it’s using (max(x0) - min(x0)) / 4, and I know that’s not working for my data.

Thank you for any suggestions,
Megan

2

Hello @mschendel and welcome to the forum!

Is there no proper EOG channel included in your data? Which MEG channels have you tried?

Best wishes,

Richard

3

Hi Richard,
At our site, we include proper EOG channels, but I’m working with data from another site without. We had a student go through the data and identify MEG channels with significant eyeblink; they are often 1411 and 0121, sometimes 1441, 1421, 0111, 0131, 0141. For the dataset I mentioned as an example, it was 1421, but 1411 and 0121 both have clear signal on them as well.

Thanks,
Megan

4

Thanks, @mschendel! Would you be able to share one of those files with me so I can take a closer look? You can send me a private message here, or reach me via email at richard.hoechenberger@gmail.com

5

Hi Richarc, I am going to try to email some data, we’ll see if that works.
In the meantime, it would be helpful to know some suggested values for neuromag magnetometers for the thresh value.
Thanks!
Megan

6

Hi @mschendel, I cannot give a general advice now, I’d need to look at your data :slight_smile: Thanks!

7

Hi Richard, I tried to email you with some data did you get it?
Megan

8

Hello @mschendel, no, unfortunately I have not received anything :frowning: I think the easiest way to share the file would be via Dropbox or Google Drive, would that work for you?

9

Hi Richard, Can you check your spam? I sent it via drive to the email address you gave. I will try again
Thanks,
Megan

10

Interesting, it’s actually listed in my Google Drive, but Google never sent me an email! Sorry about that! I will have a look tomorrow.

Best wishes,

Richard

11

Hi Richard,
How strange! Well, I’m glad you have it.
FWIW the participant is an 8-14 year old (smaller head, possibly lower in helmet than common for adults)
It is a longitudinal study with 3 visits, 5 tasks each visit, and about 90-70 subjects for each visit, so I really would like to get some automatic processing because creating projectors for every single dataset would be very cumbersome (reviewing projectors might be necessary)
Thanks!
Megan

12

Hi Richard,
Have you had any insights?
Thanks,
Megan