Dear,
When I import in database, there is a “remove DC offset : select baseline definition”, I found that there is no difference between I choose a time range or I didn’t select the option, and there is no differenece between I choose different time ranges especially in EEG. (I did MEG/EEG simultaneously). Is it a serious bug or is there anything wrong with my selection?
It is a serious problem for me. I wish your reply as soon as possible ! Thank you very much
YANG
Dear Yang,
What file format are you using?
What menu are you using to import your data into Brainstorm?
Francois
Hi Francois,
The file format is " .fif " (Elekta Neuromag MEG System).
I chose “link to raw file”, and selected “import in database”, and then I checked the “Remove DC offset” option, together with “Time range”.
Thank you for your reply! Lookin forwoard to your solution!
Yang
Hi Yang,
Sorry for the response delay, I was away last week.
I could not reproduce the behavior you are observing: when importing recordings from a FIF file, the recordings saved in the database are always different when I select different options (no DC removal, DC removal from baseline, DC removal from all the recordings).
Can you send me one of your files, together with a screen capture of the window with the import options?
It will really help me with the debugging of this problem.
You can upload your FIF file somewhere (eg. dropbox) and send me the link in a separate email (to send me an email, click on my user name on this forum > Send email).
Thanks,
Francois
Dear Yang,
Indeed, the Dropbox website is apparently blocked in China…
I checked the problem you were describing here with your file.
I tested five configurations: No DC correction, DC:All, DC:-200,0, DC:-200,-100, DC:-200,-190.
Conclusion: they are all different, the DC removal works perfectly.
Have you just looked at the traces and had the feeling they were the same or have you actually tested it?
If you import your trials with similar baselines (eg. [-200,0] vs [-200,-50]), average them, and look at the four signals, they will look very similar because the baseline is quite flat.
But it you export them to your Matlab workspace and them compare them formally, you would see that they are actually different. Another way to observe the difference easily is to maximize the figures and display them alternatively.
I also have a few remarks about the screen captures you sent me:
You have only one SSP category calculated for all the sensor types.
I would recommend you process separately the magnetometers and gradiometers, as illustrated in this tutorial:
http://neuroimage.usc.edu/brainstorm/Tutorials/TutMindNeuromag
You also have to look carefully at what the blink detection does. It detects 162 blinks, which are nicely matching all the peaks on the channel EOG062. But if you look at the topography at those blinks (with a low-pass filter at 20Hz), lots of those “blinks” do not show ANY ocular activity.
Therefore the SSP calculation performs a really bad job at identifying one clear component that captures only the blink. For the blinks, you should observe one component with a % value much higher than the others (eg. component 1: 15%, component 2: 6%).
I suspect this is because you are running the tSSS algorithm first. This gets rid of lots of the eye contamination, but not all of it…
You should spend more time on the cleaning of your blinks, maybe using the process “Compute SSP: Generic”.
Cheers,
Francois
Hi Francois,
Thank you very much for your advice!! I once exported data into Matllab workspace and tested them, I found they were same. Anyway, I will have another try. The version I used is 3.1.130925. Is this version sufficient to process as you mentioned?
Yang
Hi Yang,
No, this version is quite old. Please update to today’s version (menu Help > Update Brainstorm)
Francois