noise covariance matrix

✉️ Mailing List Archive (read-only)
1

External Email - Use Caution

Hi,
I am analyzing auditory evoked responses (N100).
I calculated NCM for pre-stimulus activity (500 ms long epochs).
For the ERPs I used filter 0.5-20, and I applied the same for the NCM
epochs. The dSPM was totally messy. Then I tried to calculate the NCM
without filters, and, surprisingly, the N100 was localized perfectly to the
auditory cortex.
Is it necessary to use the same filter before the NCM calculations?
Thank you,
Laura
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20200325/7eae41e7/attachment.html

2

External Email - Use Caution

I forgot to say, that I used filter 0.5-20 Hz for the averaged ERPs, I
didn't filter data before averaging. But the NCM is calculated from epochs,
so may be it is incorrect to filter epochs?

3

External Email - Use Caution

Hi Laura,

yes, for standard use cases, it is strongly recommended to apply the same signal processing to the data from which you compute the noise covariance and your signal of interest: filtering, ICA/SSP/SSS, etc.
Did you filter the continuous time series at raw level?

Best wishes,
Denis