RuntimeError: No clean segment found. Please consider updating your
rejection thresholds
It means that your reject parameters are too strict. You want to make sure
that data segments contaminated by strong environmental magnetic fields are
ignored, but you don't want to reject ECG or EOG since these need to be
kept in order to find corresponding components.
Thks, I supposed it was related to the reject dict and
I wonder if I could get this thresholds by inspecting the data
but with raw.plot() I didn't see the MEG field amplitudes o.O
I see where you want to go. In my experience the values I recommended to
you work pretty well on many data. Just take a look at average EOG/ECG
artefact scales and then add some room above to make sure you don't reject
data because of biological artifacts you intend to regress out using ICA.
Have a look at the last section here: http://martinos.org/mne/stable/auto_tutorials/plot_artifacts_detection.html
In my experience your values are a bit low, we typically use those for
rejection of EOG/ECG. So this is not what you want because your ICA is
supposed to learn EOG/ECG pattern by exposure to their corresponding
topographies.