I am processing MEG data for Time-Frequency Analysis. To avoid edge effects, I am using a wide time window (tmin=-1.0, tmax=2.0), even though my actual Region of Interest (ROI) is smaller (e.g., -0.2 to 0.8s).
I noticed that if I use the default rejection behavior (checking the full epoch length), I lose many trials due to artifacts occurring at the far edges (e.g., at -0.9s or 1.8s), which are outside my analysis window.
My Question: Is it considered best practice to use reject_tmin and reject_tmax to restrict artifact rejection to a narrower window (e.g., reject_tmin=-0.5, reject_tmax=1.0)?
I want to verify if it is safe to keep epochs that are clean within the ROI but noisy at the edges, specifically when the ultimate goal is TFR analysis.
welcome to the forum. This is a good question and I don’t think there is clear right or wrong approach here.
I would suggest to remove noisy epochs based on the whole time window to avoid artefacts leaking into the ROI time window.
Are the artefacts you observe random or do you see e.g. eyeblinks happening at the beginning of each epoch?
I would try to get rid of the artefacts as best as you can and then run the TFR on the clean, long epochs before cropping the TFR within your desired time window.