[Mne_analysis] Event file old vs. new format

Alexandre Gramfort gramfort at nmr.mgh.harvard.edu
Thu Sep 29 09:32:12 EDT 2011
Search archives:

Hello Linda,

I can send you the mne C code but let me try to suggest something.
It seems to me that your choice is binary either offset the time sample
in the eve file with raw.first_samp or not. The output of this experiment
should tell you easily if you do the right thing. Regarding the comparison
with the MNE C code, be careful that MNE discards bad epochs on the
fly so you'll need to pass only good events to have a real comparison.

if you feel like it, I think this example:


is doing pretty much what you want to do.

let me know if you need help to give it a try.


On Wed, Sep 28, 2011 at 11:53 PM, Linda Moya <lhmoya at cmu.edu> wrote:
> Hello Alex,
> Thank you for your very quick response. I think my question can be
> answered by looking at the code for mne_process_raw (which I gather is
> written in C, since I see no associated Matlab code, and it is an
> executable).
> The mne_process_raw program while executing at some point puts out:
> ...
> Data skip of 18000 samples in the beginning
> ...
> Event file is in the old format. Taking into account offset of 18000 samples
> Read 118 events from valid_lh_target_run1_inp.eve
> ...
> where valid_lh_target_run1_inp.eve is one of the event files I created
> (for the valid condition, left hemfield targets for the first run). So
> yes, I do think I mean the raw.first_samp field.
> The reason I want to know the exact answer to this question, is that I
> both use mne_process_raw to average the MEG time-series data, AND I use
> fiff_read_setup_read_raw and fiff_read_raw_segment to average the file
> myself for the following reason:
> 1) I re-reference the EEG data and then average it myself
> 2) I conduct a wavelet analysis on each trial before averaging for both
> the MEG and EEG data.
> I want to be sure that I am averaging it the same way (taking the initial
> skip into account) as mne_process_raw after I do these manipulations of
> the data for EEG, and for the wavelet analysis for both MEG and EEG, when
> I am doing the averaging myself. In all cases I use the same event files.
> I have read the MNE manual for fiff_read_setup_read_raw and
> fiff_read_raw_segment, but the info on these routines is cryptic and it is
> unfortunately not clear to me exactly how to use these routines to
> correctly account for the initial skip, in the same way as mne_browse_raw.
> Again in both cases I use the same event files.
> The answer to my question could probably be answered by looking at the C
> code for mne_process_raw around the place where it outputs the statement
> "Event file is in the old format. Taking into account offset ..." etc.
> Somewhere in the vicinity would exist the code to take that into account.
> If you don't want to read the code, would it be possible to send the C
> code text file to me? I don't plan to do anything with it other than read
> the code in the vicinity of that statement.
> Please let me know how I can make my question more clear so that I can get
> the answer that I need.
> Thank you for your consideration.
> Best Regards,
> -- Linda Moya
> On Wed, September 28, 2011 9:30 pm, Alexandre Gramfort wrote:
> | Hello Linda,
> |
> | I've never had to deal with old event file format but I'll try to answer.
> |
> |> In using mne_process_raw to epoch and average across trials in a given
> |> condition (output is a *.fif file for the given condition), the log output
> |> states that my event file is in the the old format. So it says it is
> |> taking into account the offset.
> |
> | by offset do you mean the raw.first_samp field?
> |
> |> First, how exactly does it take into account the offset? In other words
> |> how does the sample start time ultimately used to index the raw fif file
> |> differ from the sample start time specified in the event file?
> |
> | I guess in one case the first time point is 0 in the other it
> corresponds to
> | offset.
> |
> |> Second, how does the new format differ from the old format? The format I
> |> am using is:
> |> <sample>  <time>  <from> <to>
> |>
> |> where <sample> is a number, <time> turns out to always be sample/1000
> |> <from> is 0 and <to> is the specific condition number.
> |
> | no clue.
> |
> |> I am creating my own event files, so that I can cut the data many
> |> different ways not necessarily specified by the triggers in the raw fif
> |> file, and want to make sure that the format is accurate.
> |
> | I would try with a simple dataset to see if the approach produces a
> | nice evoked response.
> |
> | sorry for not being able to help more
> |
> | Alex
> |
> |
> | The information in this e-mail is intended only for the person to whom
> it is
> | addressed. If you believe this e-mail was sent to you in error and the
> e-mail
> | contains patient information, please contact the Partners Compliance
> HelpLine at
> | http://www.partners.org/complianceline . If the e-mail was sent to you
> in error
> | but does not contain patient information, please contact the sender and
> properly
> | dispose of the e-mail.
> |
> |
> |
> --
> Linda Moya, Ph.D.
> Electrical and Computer Engineering
> Center for the Neural Basis of Cognition
> Carnegie Mellon University
> http://www.andrew.cmu.edu/user/lhmoya/

More information about the Mne_analysis mailing list