help
help > RE: Error in conn_process (line 1398)
Jun 9, 2014 02:06 AM | Alfonso Nieto-Castanon - Boston University
RE: Error in conn_process (line 1398)
Hi Roger,
Regarding your conn_*.mat file, the only (but perhaps unrelated) issue that I see in your setup definitions is that the condition onset and duration field values seem to be entered in ms (but CONN expects them to be defined in seconds), so most of your condition blocks are going to be outside of the range of your scans.
And regarding the "invalid" trials, you can remove those from consideration by adding one additional first-level covariate where you define the invalid trials, and then adding this covariate in the Preprocessing step to the confounds list (the process is the same as in "scrubbing", used to eliminate outlier scans detected, for example, using art). In order to do this you need to have (or create) a series of files (one file for each subject&session) identifying the invalid trials. These files can be defined as .mat or .txt files; they should have as many rows as scans (for each subject/session), and as many columns as invalid trials. Values are 1 or 0; with a 1 in the row i and column j identifying the j-th invalid trial as occurring in scan i (you may have one or multiple scans/rows associated with each "trial"/column).
Hope this helps
Alfonso
Originally posted by Roger Beaty:
Regarding your conn_*.mat file, the only (but perhaps unrelated) issue that I see in your setup definitions is that the condition onset and duration field values seem to be entered in ms (but CONN expects them to be defined in seconds), so most of your condition blocks are going to be outside of the range of your scans.
And regarding the "invalid" trials, you can remove those from consideration by adding one additional first-level covariate where you define the invalid trials, and then adding this covariate in the Preprocessing step to the confounds list (the process is the same as in "scrubbing", used to eliminate outlier scans detected, for example, using art). In order to do this you need to have (or create) a series of files (one file for each subject&session) identifying the invalid trials. These files can be defined as .mat or .txt files; they should have as many rows as scans (for each subject/session), and as many columns as invalid trials. Values are 1 or 0; with a 1 in the row i and column j identifying the j-th invalid trial as occurring in scan i (you may have one or multiple scans/rows associated with each "trial"/column).
Hope this helps
Alfonso
Originally posted by Roger Beaty:
Hi Alfonso,
Thanks for the advice. In checking the Setup window, all ROIs appear to have been defined for all subjects. Also, the ROI files were not entered in the confounds list. I'm now re-running the analyses from scratch, and will keep you posted on the outcome. In the meantime, my conn_*.mat file is attached.
I have another question regarding First Level analysis. There are some "invalid" experimental trials that I would like to exclude from the model (i.e., trials which participants did not produce a valid behavioral response). Is there some way to dummy code valid and invalid trials in Conn?
Roger
Thanks for the advice. In checking the Setup window, all ROIs appear to have been defined for all subjects. Also, the ROI files were not entered in the confounds list. I'm now re-running the analyses from scratch, and will keep you posted on the outcome. In the meantime, my conn_*.mat file is attached.
I have another question regarding First Level analysis. There are some "invalid" experimental trials that I would like to exclude from the model (i.e., trials which participants did not produce a valid behavioral response). Is there some way to dummy code valid and invalid trials in Conn?
Roger
Threaded View
| Title | Author | Date |
|---|---|---|
| Roger Beaty | Jun 7, 2014 | |
| Alfonso Nieto-Castanon | Jun 8, 2014 | |
| Roger Beaty | Jun 8, 2014 | |
| Alfonso Nieto-Castanon | Jun 9, 2014 | |
| Roger Beaty | Jun 9, 2014 | |