Problem loading .study files

Help
2013-09-26
2013-10-12
  • Bruno G. M.

    Bruno G. M. - 2013-09-26

    Hello!

    I've recently downloaded EP Toolkit mainly to use the wonderful PCA functions for it, but I'm having a hard time reading .study files from EEGlab. I set it to read EEGlab (.set/.study) format, average or grand average for File Type and uncheck the Single Cell Files.

    When I press the read button, I get 2 error windows:

    Defective file will not be loaded.
    and
    File not read successfully. Aborting effort to merge data files.

    In the command window, I get the following warning/error text:

    Warning: No 'value' field in the events structure.
    EEGlab data files should have both a 'value' field
    to denote the generic type of event, as in 'trigger', and a 'type' field
    to denote the nature of this generic event, as in the condition of the experiment.
    Note also that this is the reverse of the FieldTrip convention.
    Read in 26 cells and 1 subject.
    Note: No reference channel was found.
    The name of the experiment is:
    The pre-stimulus period of the data is: 1000 msec.
    The sample rate of 600 is not consistent with the sample sizes which are 1.6667.


    Defective file will not be loaded.



    File not read successfully. Aborting effort to merge data files.


    We already did the pre-processing using EEGlab functions.

    I hope someone can help me about it. I'll be waiting. Thanks in advance!

    Bruno G. Massa.

     
  • Joe Dien

    Joe Dien - 2013-09-26

    Hi Bruno,
    it sounds like you have a couple problems. The problem with the events structure is that EEGlab doesn't seem to be consistent about following its file structure convention so it's hard to interpret them. If you can send me a copy of your file, I can try to adapt the EP Toolkit to handle your situation. The other problem is that your sampling rate seems to be incorrect, or at least inconsistent with the rest of the data. Are you actually taking 600 samples per second? That's a very unusual sampling rate.

     
    • Bruno G. M.

      Bruno G. M. - 2013-10-02

      Hey, Joe!

      I've done some testing and I found out that the problem was after I used EEGlab to divide the trials in epochs.

      Here are the comparison between the non-epoched and epoched command window log:

      Non-Epoched:

      Warning: No 'value' field in the events structure.
      EEGlab data files should have both a 'value' field
      to denote the generic type of event, as in 'trigger', and a 'type' field
      to denote the nature of this generic event, as in the condition of the experiment.
      Note also that this is the reverse of the FieldTrip convention.
      Read in 1 cell and 1 subject.
      Note: No reference channel was found.
      The name of the experiment is:
      The pre-stimulus period of the data is: 0 msec.

      Epoched:

      Warning: No 'value' field in the events structure.
      EEGlab data files should have both a 'value' field
      to denote the generic type of event, as in 'trigger', and a 'type' field
      to denote the nature of this generic event, as in the condition of the experiment.
      Note also that this is the reverse of the FieldTrip convention.
      Read in 100 cells and 1 subject.
      Note: No reference channel was found.
      The name of the experiment is:
      The pre-stimulus period of the data is: 1000 msec.
      The sample rate of 600 is not consistent with the sample sizes which are 1.6667.


      Defective file will not be loaded.


      As you can see, the only difference between the reports is the pre-stimulus period of the data and the sample rate inconsistency.

      I used the following function in EEGlab (Tools > Extract Epochs from the Menu) to set the epochs. As you can see, we used 2 triggers (110 and 111) for reference, a time window from -1 to 3 seconds and a custom variable with the wanted name for it (epoched.save). The same result is obtained by doing it manually.

      EEG = pop_epoch( EEG, {'110' '111'}, [-1 3], 'newname', epoched.save, 'epochinfo', >'yes');
      EEG = pop_rmbase( EEG, [-1000 0]);

      I'll upload non-epoched and epoched sample files soon, and post it here. The connection where I am right now is too slow, I'll do it somewhere else. The 2 .sets and .fdts together sum up roughly 300 MB.

      I hope that helps for now. Thank you very much!

      Bruno Massa.

       
  • Joe Dien

    Joe Dien - 2013-10-02

    That would be great! Also, could you tell me how the non-epoched .set files were originally generated? I'm trying to figure out why sometimes EEGlab files have a 'value' field and sometimes not.

     
  • Bruno G. M.

    Bruno G. M. - 2013-10-03

    Sorry, I'm kinda late. There you go.
    http://www.speedyshare.com/XtedT/datasets.rar

    We use a custom built-in function in MatLab to import .peb files (that's the file extension generated by our acquisition software called AcquireEEG). We did some preprocessing like filtering, rereferencing, setting the event channel, naming it and loading the scalp map reference file (.ced).

    I think the problem is in the sample rate, since the non-epoched .set can be loaded aparently fine.

    Thanks once again!

     
  • Joe Dien

    Joe Dien - 2013-10-03

    I tried to load and got this error message:


    No data were read. The error message was:
    Error using read_eeglabdata (line 74)
    Cannot not find data file: teste_triggershareq_filtered0545reref_epochs.fdt


    Does this sound familiar to you? Have you tried loading in the files you are sending me?

     
  • Joe Dien

    Joe Dien - 2013-10-03

    I worked around the problem by changing the name of the .fdt file back to the original name. So another issue I'm seeing is that the "type" field from the ced file is a set of numbers rather than labels like "FID" and "EEG".

     
  • Joe Dien

    Joe Dien - 2013-10-03

    Okay, the problem was that my code was not handling rounding errors when comparing the sampling rate and the sample labels. I've just sent you a link to a 2.40beta copy that should fix this problem. Please let me know if it works for you.

     
  • Bruno G. M.

    Bruno G. M. - 2013-10-04

    I'll test it tomorrow and tell you the results. Thanks a lot in advance!

    And yeah, I didn't know changing the file name would mess things around. I just wanted to make the files names shorter.

     
  • Bruno G. M.

    Bruno G. M. - 2013-10-10

    Once again I'm late, I'm sorry for that. We were in a hurry to finish something else, so I couldn't focus on testing. Well, it seems like the .set reading was fixed! However, the .study reading gave me another error:

    Warning: No 'value' field in the events structure.
    EEGlab data files should have both a 'value' field
    to denote the generic type of event, as in 'trigger', and a 'type' field
    to denote the nature of this generic event, as in the condition of the experiment.
    Note also that this is the reverse of the FieldTrip convention.
    Read in 26 cells and 1 subject.
    Warning: the 'type' field from the CED file contained numbers rather than labels like EEG >or REF.
    Note: No reference channel was found.
    The name of the experiment is:
    The pre-stimulus period of the data is: 1000 msec.
    Warning: No 'value' field in the events structure.
    EEGlab data files should have both a 'value' field
    to denote the generic type of event, as in 'trigger', and a 'type' field
    to denote the nature of this generic event, as in the condition of the experiment.
    Note also that this is the reverse of the FieldTrip convention.
    Read in 41 cells and 1 subject.
    Warning: the 'type' field from the CED file contained numbers rather than labels like EEG >or REF.
    Note: No reference channel was found.
    Warning: No 'value' field in the events structure.
    EEGlab data files should have both a 'value' field
    to denote the generic type of event, as in 'trigger', and a 'type' field
    to denote the nature of this generic event, as in the condition of the experiment.
    Note also that this is the reverse of the FieldTrip convention.
    Read in 39 cells and 1 subject.
    Warning: the 'type' field from the CED file contained numbers rather than labels like EEG >or REF.
    Note: No reference channel was found.


    Error: The file C:\Users\Micro\Documents\MATLAB\STUDY\ASq70.set has different channels >from the first file.


    Something went wrong. I guess we can still work our way out of it by loading the .set's individually, but I don't know if you wanna solve that problem as well. Thank you for the fix to the .set loading!

    Bruno Massa.

     
  • Joe Dien

    Joe Dien - 2013-10-10

    Glad to hear the .set problem is solved at least! You'll need to send me the files so that I can look them over, both the first .set file, the second one that set off the error, and the .study file.

     
  • Joe Dien

    Joe Dien - 2013-10-12

    also, could you take a look at the channels in the two files using the Edit function? If they do indeed have different channels (or at least channel names), then this is a problem with how the files were created and thus not something I can help with.

     

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:

JavaScript is required for this form.





No, thanks