search

Mensen / ept_TFCE-matlab

Advanced EEG Statistics
28 stars 9 forks source link

Missing channels? #22

Open JD-Zhu opened 6 years ago

JD-Zhu commented 6 years ago

Dear Armand,

Thank you for developing this great tool!

I would like to run this on my ERF data, and was wondering if it can readily take in data with missing channels? I did some manual rejection of bad channels during preprocessing (different channels were rejected for different participants). Could I simply fill in those missing channels with NaNs before passing the data into TFCE, or would I need to somehow replace those channels (e.g. by interpolation)?

Many thanks, Judy

Mensen commented 6 years ago

Dear Judy, Unfortunately you would need to replace those channels through interpolation for each participant. The statistics handle all channels (and time points etc) in one whole batch, rather than in some sort of loop checking for missing values etc, so the data needs to be consistent. There are also other difficulties with how to then deal with clusters coming from statistical values derived from different participant numbers which are not ideal to work with either. So I agree that working with interpolated data (sometimes), is not ideal, the tradeoff to other options still favours this approach (in my opinion). Good luck!

JD-Zhu commented 6 years ago

Thanks Armand - I'll go with channel interpolation then!