[Eeglablist] ICLabel for eyes closed resting state data

[Amy Tong]

Hi all,

I'm working with eyes-closed resting state data, and using ICLabel to reject eye components. I've seen that a threshold of 0.8 probability or higher is generally recommended, but I'm wondering how accurate ICLabel would be for identifying eye components during an eyes-closed paradigm and if a lower threshold (e.g. 0.7) would make sense, as I would assume that most training data was eyes-open. Has anyone got any experience of this, or any thoughts?

Best wishes,
Amy