mne_process_raw problem

Hi,

I am having a problem in one subject when trying to average raw files
(and computing a noise covariance matrix) with mne_process_raw. I sent
an mail about this a while ago but did not hear back and all my attempts
to solve this problem myself failed.

The error message says:
"Event file is in the old format. Taking into account offset of 135000
samples
Read 714 events from SBT_MEG_005_run1_converted.eve
Computing...
Too few samples (required : 1885 got : 0)"

My understanding of this error message is that mne does not find events
listed in the event file that match events in the raw file. However, the
events listed in the event file do match events on the trigger channel
(STI 014). Played around with the event file using non-modified event
files and event files in the old and new format did not lead to any
success, nor did excluding runs separately or excluding EEG from the raw
files (the EEG looked not nice because the subject was sweating a lot).
Running the old mne version did not change anything either.
I have run the same command in other subjects without problems.

Any help with this is greatly appreciated!
thanks, Sarah

Hi Sarah,

I believe this shows the data is being rejected for those samples. You
should look at the rejection criteria (should be identical to averaging),
next check the log file. Send another e-mail if those don't look promising

Dan

Hi Sarah,

One idea is to check in your off-line averaging description file the
*Reject parameters since maybe they are too high and thus all your epochs
are being rejected.

Before doing the offline averaging you should do it online
(mne_browse_raw), changing these values in Adjust/Averaging preferences to
be sure you will get some.

Hope this helps,

-- Andr?s

Hi,

I am having a problem in one subject when trying to average raw files

(and computing a noise covariance matrix) with mne_process_raw. I sent an
mail about this a while ago but did not hear back and all my attempts to
solve this problem myself failed.

The error message says:
"Event file is in the old format. Taking into account offset of 135000

samples

Read 714 events from SBT_MEG_005_run1_converted.eve
Computing...
Too few samples (required : 1885 got : 0)"

My understanding of this error message is that mne does not find events

listed in the event file that match events in the raw file. However, the
events listed in the event file do match events on the trigger channel
(STI 014). Played around with the event file using non-modified event
files and event files in the old and new format did not lead to any
success, nor did excluding runs separately or excluding EEG from the raw
files (the EEG looked not nice because the subject was sweating a lot).
Running the old mne version did not change anything either.

Hi Sarah,

Also, it could be possible that you are getting this error message because
there are too few samples to calculate the noise cov matrix.

I am guessing if you invoke mne_process_raw ONLY to calculate the grand
average, it should execute OK.

~N&ita