1/f compensation for selecting epochs with low frequency components

Dear all,

I am working on a method to select epochs made from a large dataset that might contain focal slowing. After preprocessing and dropping epochs with mne-faster, the most obvious next step is selecting epochs based on a frequency criterion.

At the moment, I integrate the PSD of each epoch in the different frequency bands, multiplying the obtained values with their respective frequencies. This approach sounds fair if we’re looking for a difference in low-frequency power between epochs.

My question is if this approach makes scientific sense or if someone has experience with it. I used to use Brainstorm, which includes a 1/f compensation for its time-frequency plots, but I could not find such compensation in MNE by itself.

Kind regards,

Hello Steven,

there is no 1/f compensation implemented in MNE yet. However you could use the fooof toolbox which is also python based and allows for 1/f estimation.