Removing trials from script

1

Hello Francois,

I made a script to do block averaging. I first import the block in the database and mark the trial as bad depending on some criteria I have. This is working. However, when I then call baseline normalization. it's also considering the bad trials and not only the good ones. Is there something I am missing?

sTrials = bst_process('CallProcess', 'process_import_data_event', files, , ...
'subjectname', sub_name{:}, ...
'condition', '', ...
'eventname', 'Hard, Easy', ...
'timewindow', , ...
'epochtime', t_windows, ...
'createcond', 0, ...
'ignoreshort', 0, ...
'usectfcomp', 0, ...
'usessp', 0, ...
'freq', , ...
'baseline', );

for iTrial = 1:length(sTrials)
        process_detectbad('SetTrialStatus', sTrials(iTrial).FileName, ~trial_status(iTrial));
end   

% Process: DC offset correction: [-10.000s,0.000s]
sTrialsNorm = bst_process('CallProcess', 'process_baseline_norm', sTrials, [], ...
    'baseline',    t_baseline, ...
    'sensortypes', 'NIRS', ...
    'method',      'bl', ...  % DC offset correction:    x_std = x - μ
    'overwrite',   0);

% here sTrialsNorm also contains bad trials

Regards,
Edouard

2

The bad trials are excluded from the trial selection, not from bst_process('CallProcess',...).
It means that you'd need to use another search or select process to grab again the files from the database:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting#Starting_a_new_script
https://neuroimage.usc.edu/brainstorm/Tutorials/PipelineEditor#Search_Database

It's probably easier to call the process process_baseline_norm with only the trials you want: sTrials(trial_status)

1 Like
3

thanks. it's working :slight_smile:

4

You may also find this useful for interfacing to mark bad trials: https://github.com/nichrishayes/ArtifactScanTool

5

@Nichrishayes Wouldn't you be interested in implementing this as a Brainstorm process? :slight_smile: https://neuroimage.usc.edu/brainstorm/Tutorials/TutUserProcess

6

@Francois, briefly yes :blush:. That'd probably be a good end solution. Not sure I'll be able to get to it in the near-term :frowning_face:. @aiwiesman may be working on that end already

7

Let me know if you need help with further integration, new features and new contributors are always welcome!

8

Hi Nick and Francois,

Yes indeed, developing a semi-automated approach to bad epoch/trial identification is on my to-do list, with help from @Raymundo.Cassani, but has unfortunately gotten a bit buried by other commitments. I'll revisit with Raymundo et al. soon and update here when we have a plan in place.

Alex