PSD differences between MNE and EEGLAB

[const (Kun CHEN)]

Dear all,

I recently compared the spectrum (PSD) results of my SEEG data between EEGLAB and MNE. I noticed that there are minor differences between them.

Environment

• MNE version: 1.7.1
• OS: Windows 11

Relevant code

I use the default setup for PSD plots in both MNE and EEGLAB.

# python
import mne
raw = mne.io.read_raw_edf(sub_seeg_path, preload=True)
raw.resample(500, n_jobs=32)
raw.compute_psd(fmax=200, n_jobs=32).plot()

% matlab
EEG = pop_biosig('D:\OneDrive\working_data\sEEG_data\seeg\xxx\seeg\xxx.edf');
EEG = pop_resample( EEG, 500);
figure; pop_spectopo(EEG, 1, [0  2252998], 'EEG' , 'freqrange',[0.1 200],'electrodes','off');'

The MNE PSD plot shows multiple spikes (highlighted within the orange rectangles), while the EEGLAB plot does not. I would like to know why they are different and what these spikes mean for my data.

[agramfort (Alexandre Gramfort)]

the default frequency resolution of MNE is higher. You should reduce the value of the n_fft parameter.

eg

raw.compute_psd(n_fft=128)

Alex

[const (Kun CHEN)]

Thanks you Alex. I checked the document about n_fft in compute_psd and psd_array_welch, but the description is quite simple. Can you tell me more about its usage/function?

[martin]

Have a look at the **method_kw section of compute_psd. You can pass n_fft=X with X matching what EEGLAB uses.

[const (Kun CHEN)]

thank you!