Trial-specific bad channels

[dweisholtz]

I have marked epileptic discharges as channel-specific bad segments in a raw file. When I import my data, the channels marked bad vary from trial to trial, but I have no completely bad trials (at least some channels are good in every trial). I have 100 trials. When I compute the average across trials (ERP), my understanding is that although all 100 trials were incorporated into the average, the bad channels were excluded when appropriate.

My question: Is it possible to tell post-hoc how many trials went into each channel's calculation? Or do I need to simply count up how many bad trials there were for each channel individually?

Many thanks!

Best,
Dan

[Raymundo.Cassani]

Hi @dweisholtz,

@dweisholtz

When I compute the average across trials (ERP), my understanding is that although all 100 trials were incorporated into the average, the bad channels were excluded when appropriate.

Yes, this is correct. For each epoch, channels that overlap with a channel-specific bad segment are ignored for the average.

@dweisholtz

My question: Is it possible to tell post-hoc how many trials went into each channel's calculation? Or do I need to simply count up how many bad trials there were for each channel individually?

Yes, it is possible with a simple script. The idea is to open the ChannelFlag for each epoch and count how many times channels appear as Good.

You can generate a script to generate the average across epochs.
https://neuroimage.usc.edu/brainstorm/Tutorials/PipelineEditor#Saving_a_pipeline

Then modify it to have the channels counted as in this script:
https://gist.github.com/rcassani/bd347e6b2a8707b98052072b0eb30f00

Best,
Raymundo