GFP vs Cluster

[ivano7777]

Hello,

I tried the tool following the tutorial, but applying it to MEG, instead of EEG (of course modifying a few details). I liked the individual microstates that I computed, after selecting my "meg_good_segments" this way:

epochs = mne.make_fixed_length_epochs(meg_good_segments, duration=2)
#I tried without, results are similar. but don;t use too short epochs. ModK = ModKMeans(n_clusters=n_microstates, random_state=42) ModK.fit(epochs, n_jobs=4, picks = 'meg') ModK.rename_clusters(new_names=["A", "B", "C", "D"]) ModK.plot(show_gradient=True)

Then, always according to the tutorial, for group analysis (whatever we want to use clusters or GFP) we need to compute the individual gfp. I did it, and lookes at the microstates using the gfp, and they are weird.

Questions:

1. Is it normal that they are so different? (GFP vs Cluster?)
2. What does the software do when it doesn't use the GFP for the fit? I am not sure I got it.

Thanks