help
help > RE: Removing specific runs or conditions for a single subject
May 18, 2022 06:05 PM | rogojin
RE: Removing specific runs or conditions for a single subject
Hello,
Just wanted to follow up on my previous post. Following this previous answer on the forum re: excluding invalid trials (https://www.nitrc.org/forum/message.php?...), I was wondering if the setup below would be correct.
I have 5 sessions per subject, with 8 conditions per session. Each session has 473 volumes/scans. For simplicity's sake, I'm going to use an example where each session is 10 volumes long (instead of my actual 473).
1) If for example one of my subjects messed up condition #1 in one session and this condition happens to occur during volumes 4-7, would the 1st level covariate file need to be structured as follows:
0 0 0 0
0 0 0 0
0 0 0 0
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
0 0 0 0
0 0 0 0
0 0 0 0
Above, volumes 4-7 contain a value of 1 and presumably would be excluded as confounds to be regressed out of the model. The 6 of 10 rows (i.e., volumes 1-3 and volumes 8-10) that contain only 0s would be included as valid. So, in my real data when I have a condition to exclude, I would have to look at the run/session timing file to figure out which of the 473 volumes that condition took place in.
Is this correct?
2) If the above is correct, why do these 1s have to occur in separate columns? For example, why would the covariate file not just be:
0
0
0
1
1
1
1
0
0
0
If I know that condition 1 that I want to exclude occurs in volumes/scans #4-7?
3) If I want to exclude an entire session/run that was invalid (for example, the subject was sleepy and not focused on the task), would this be the correct format:
1 0 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0
0 0 0 0 1 0 0 0 0 0
0 0 0 0 0 1 0 0 0 0
0 0 0 0 0 0 1 0 0 0
0 0 0 0 0 0 0 1 0 0
0 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 0 1
Or, in my real data, a 473x473 matrix with 1s along the diagonal?
4) Can I just include these files for sessions where there were invalid conditions I want to exclude? I.e., for a sessions where I have nothing to exclude, is no file needed?
Thank you!
Alica
Originally posted by rogojin:
Just wanted to follow up on my previous post. Following this previous answer on the forum re: excluding invalid trials (https://www.nitrc.org/forum/message.php?...), I was wondering if the setup below would be correct.
I have 5 sessions per subject, with 8 conditions per session. Each session has 473 volumes/scans. For simplicity's sake, I'm going to use an example where each session is 10 volumes long (instead of my actual 473).
1) If for example one of my subjects messed up condition #1 in one session and this condition happens to occur during volumes 4-7, would the 1st level covariate file need to be structured as follows:
0 0 0 0
0 0 0 0
0 0 0 0
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
0 0 0 0
0 0 0 0
0 0 0 0
Above, volumes 4-7 contain a value of 1 and presumably would be excluded as confounds to be regressed out of the model. The 6 of 10 rows (i.e., volumes 1-3 and volumes 8-10) that contain only 0s would be included as valid. So, in my real data when I have a condition to exclude, I would have to look at the run/session timing file to figure out which of the 473 volumes that condition took place in.
Is this correct?
2) If the above is correct, why do these 1s have to occur in separate columns? For example, why would the covariate file not just be:
0
0
0
1
1
1
1
0
0
0
If I know that condition 1 that I want to exclude occurs in volumes/scans #4-7?
3) If I want to exclude an entire session/run that was invalid (for example, the subject was sleepy and not focused on the task), would this be the correct format:
1 0 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0
0 0 0 0 1 0 0 0 0 0
0 0 0 0 0 1 0 0 0 0
0 0 0 0 0 0 1 0 0 0
0 0 0 0 0 0 0 1 0 0
0 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 0 1
Or, in my real data, a 473x473 matrix with 1s along the diagonal?
4) Can I just include these files for sessions where there were invalid conditions I want to exclude? I.e., for a sessions where I have nothing to exclude, is no file needed?
Thank you!
Alica
Originally posted by rogojin:
Hello,
I have 5 experimental runs per subject, with 8 conditions per run. Some of the runs are not usable (i.e., subject was sleepy) and I'd like to remove them - am I able to just delete it in the setup tab? Or would I need to include all 5 runs for each subject, and create some sort of covariate file for removing them in the final analysis?
Similarly, some subjects have conditions I need to remove in a specific run (e.g., I need to remove condition 3 only in run 5 for subject 10). How can I go about removing these from final analysis? Would I need to code them somehow in a covariate file that gets loaded into a list of confounds to be regressed out at the preprocessing stage?
Thank you!
All the best,
Alica
I have 5 experimental runs per subject, with 8 conditions per run. Some of the runs are not usable (i.e., subject was sleepy) and I'd like to remove them - am I able to just delete it in the setup tab? Or would I need to include all 5 runs for each subject, and create some sort of covariate file for removing them in the final analysis?
Similarly, some subjects have conditions I need to remove in a specific run (e.g., I need to remove condition 3 only in run 5 for subject 10). How can I go about removing these from final analysis? Would I need to code them somehow in a covariate file that gets loaded into a list of confounds to be regressed out at the preprocessing stage?
Thank you!
All the best,
Alica
Threaded View
| Title | Author | Date |
|---|---|---|
| rogojin | May 16, 2022 | |
| rogojin | May 18, 2022 | |
| Isabelle Faillenot | Dec 10, 2024 | |