how to process (ica) data from multiple trials of a single subject

I would like to ask you how to process data from multiple trials of a single subject

I have 8 fif files for each story, each fif consists of 10 minutes of long EEG recordings with only a start marker and an end marker.
I still need to split them into 8 for subsequent analysis after I finally do the pre-processing such as ica and filtering.

Should I merge them into one big raw or is there a more appropriate way to do this?

  • MNE version: e.g. 1.3.0
  • operating system: e.g. Ubuntu 20.04

ICA does not care about the temporal order, so you can concatenate everything and run ICA. Other processing steps such as filtering are sensitive to boundary (edge) artifacts, which might (will?) occur when you concatenate data from multiple recording sessions. You should then mark (ignore) these sections in your subsequent analyses.

Alternatively, I think it’s maybe easier to process each recording separately, especially if you are going to have to split them anyway. So maybe a good approach is to run ICA with all recordings and perform separate processing for all other steps.

1 Like

MNE does that automatically for you if you use the built-in functions for concatenation and processing.


True, but these annotations are only at the exact time points where you concatenated (duration 0). I thought that it would be necessary to ignore a little more data around these annotations. But yes, it is all built in!

1 Like