help
help > RE: 2nd level contrast
Jan 15, 2020 03:01 PM | Alfonso Nieto-Castanon - Boston University
RE: 2nd level contrast
Dear Ali,
That should work just fine (defining two conditions in Setup.Conditions -preRest and postRest- and then associating both conditions with the same/single session for the control subjects and with the two different sessions for the patients). Could you please elaborate on the problem that you are finding in the denoising step? (it is perfectly fine that you will see the "Effect of preRest" and "Effect of postRest" confounding effects listed for the control subjects even if those two effects actually model exactly the same thing in these subjects; that redundancy will be handled by the denoising regression model without a problem).
Then your second-level models would be:
1) for the "control-prePatient" comparison you would simply select the condition "preRest" in the conditions list, and the two groups ("patient" and "control") in the subject-effects list, and enter a [-1 1] between-subjects contrast (this will be a two-sample t-test analysis)
2) for the "control-postPatient" comparison you would simply select the condition "postRest" in the conditions list, and the two groups ("patient" and "control") in the subject-effects list, and enter a [-1 1] between-subjects contrast (this will also be a two-sample t-test analysis)
3) for the "prePatient-postPatient" comparison select both "preRest" and "postRest" conditions in the conditions list with a [-1 1] between-conditions contrast, and then select only the "patient" group in the subject-effects list (this will be a paired t-test analysis)
Hope this helps
Alfonso
Originally posted by Ali Bayram:
That should work just fine (defining two conditions in Setup.Conditions -preRest and postRest- and then associating both conditions with the same/single session for the control subjects and with the two different sessions for the patients). Could you please elaborate on the problem that you are finding in the denoising step? (it is perfectly fine that you will see the "Effect of preRest" and "Effect of postRest" confounding effects listed for the control subjects even if those two effects actually model exactly the same thing in these subjects; that redundancy will be handled by the denoising regression model without a problem).
Then your second-level models would be:
1) for the "control-prePatient" comparison you would simply select the condition "preRest" in the conditions list, and the two groups ("patient" and "control") in the subject-effects list, and enter a [-1 1] between-subjects contrast (this will be a two-sample t-test analysis)
2) for the "control-postPatient" comparison you would simply select the condition "postRest" in the conditions list, and the two groups ("patient" and "control") in the subject-effects list, and enter a [-1 1] between-subjects contrast (this will also be a two-sample t-test analysis)
3) for the "prePatient-postPatient" comparison select both "preRest" and "postRest" conditions in the conditions list with a [-1 1] between-conditions contrast, and then select only the "patient" group in the subject-effects list (this will be a paired t-test analysis)
Hope this helps
Alfonso
Originally posted by Ali Bayram:
Dear All,
We have 1 session for control group and 2 sessions(pre and post treatment) for Patient group.
We want to compare functional connectivities (seed to voxel) among those three dataset
(control-prePatient, control-postPatient, prePatient-postPatient).
CONN allow us to define 1 or 2 sessions for a subject (with session invariant structural image).
We defined 2 conditions: preRest & postRest for comparison in the second level anaysis.
One session of Control group defined for both conditions, while two sessions of Patient group defined separately for conditions.
But we have problem while adding effect of rest at the denosing step. There is double "Effect of rest" for control group (pre and postRest) due to the our condition definition style.
Do you have any advise for this type of comparisons?
Is there a contrast definition at the second level for comparing control(preRest) vs. patient(postRest)?
Best Regards.
Ali
We have 1 session for control group and 2 sessions(pre and post treatment) for Patient group.
We want to compare functional connectivities (seed to voxel) among those three dataset
(control-prePatient, control-postPatient, prePatient-postPatient).
CONN allow us to define 1 or 2 sessions for a subject (with session invariant structural image).
We defined 2 conditions: preRest & postRest for comparison in the second level anaysis.
One session of Control group defined for both conditions, while two sessions of Patient group defined separately for conditions.
But we have problem while adding effect of rest at the denosing step. There is double "Effect of rest" for control group (pre and postRest) due to the our condition definition style.
Do you have any advise for this type of comparisons?
Is there a contrast definition at the second level for comparing control(preRest) vs. patient(postRest)?
Best Regards.
Ali
Threaded View
| Title | Author | Date |
|---|---|---|
| Ali Bayram | Jan 15, 2020 | |
| Alfonso Nieto-Castanon | Jan 15, 2020 | |
| Ali Bayram | Jan 16, 2020 | |