How to manage bad amplitude?

Discussions
1

Hello,
I’ve used BrainVision Analyzer and I am newly learning brainstorm.
I mostly study with ERP data and think it’s easy to learn through the tutorials on this web, thank you.

My problem is that I can’t find how to manage bad amplitudes.
On the BrainVision Analyzer, I use “raw data inspection” detecting bad amplitudes and extracting the channel or doing other solutions based on the result of the calculation provided.
On brainstorm, however, I just can find bad time segments, trials or channels by visual inspections or ICA.
ICA is for artifacts removement such as eye movements… right?
I’d like to know whether it is possible to remove bad amplitudes including abnormal waves or noise…
Thank you.

Best regards,
Nayeon

1 Like
2

Hello,

We don’t offer an integrated routine that would do a fully automated cleaning at once. “Bad amplitudes” or “abnormal waves” are very subjective notions, you will not find two programs or two researchers that consider the same bad data segments. In Brainstorm you have a variety of techniques you can use to detect bad channels and bad segments of recordings, most of them are documented on the website:
http://neuroimage.usc.edu/brainstorm/Tutorials/ArtifactsDetect#Detection:_Heartbeats (detection based on ECG/EOG)
http://neuroimage.usc.edu/brainstorm/Tutorials/ArtifactsDetect#Other_detection_processes (manual detection)
http://neuroimage.usc.edu/brainstorm/Tutorials/BadSegments (automatic detection based on power changes)

By “bad amplitudes”, do you mean you want to mark as bad the channels or trials when the EEG amplitude go above a certain threshold?
You could use the process “Events > Detect events above threshold” on continuous recordings. It creates extended events. If you include the tag “bad” in the event name, all the detect segments will be ignored from the analysis ().
Otherwise you can use the process “Events > Detect events above threshold” after importing your epochs, it marks either the channels or the trials as bad.

SSP and ICA are two techniques for removing the contributions of some specific spatial components. ICA is more common and efficient in EEG analysis.
You can refer to the documentation of the EEGLAB software. Brainstorm uses directly the function runica.m from EEGLAB.

Cheers,
Francois

1 Like
3

I will study more with your advice.
Thank you very much.

Best regards,
Nayeon

4

One more question here.

I tried “Events > Detect events above threshold” and I think this is what I was looking for.
This is to determine one channel at a time, I think, but is it possible to mark all the channels above the certain threshold?

Best wishes,
Nayeon

5

No, this process is designed to run only on one channel. You can combine multiple channels using the montage syntax (eg. “F1, -0.5M1, -0.5M2”) but all the channels are summed into one signal before the detection.

Two processes work on all the channels at once to detect bad segments:

Francois

1 Like