help
help > RE: condition specific correlation maps?
Jul 4, 2015 01:07 AM | Alfonso Nieto-Castanon - Boston University
RE: condition specific correlation maps?
Hi Shounak,
Yes, that is perfectly fine. If I am interpreting correctly that is the approach described in Fair, Damien A., et al. "A method for using blocked and event-related fMRI data to study "resting state" functional connectivity." Neuroimage 35.1 (2007): 396-405.
In general, when you have a task-related design and you are not interested in obtaining condition-specific connectivity measures, you should still enter the design information in Setup.Conditions and then simply right-click on your task conditions and select the option named "move selected condition to covariate list (Fair et al. resting state analyses of task-related data)". That will remove these task conditions from your conditions list but it will also create a set of new first-level covariates that describe the task-related main effects on the BOLD signal. When you get to the Denoising step, simply enter the "Effect of #" effects into the list of confounding effects in order to regress out these effects before computing any connectivity measures. After doing these steps you can continue with first-level analyses as you would do normally for a true "resting state" dataset, just analyzing the "rest"-condition connectivity. These steps are considered better than simply disregarding the condition information (and treating your task-related data as if it was a resting-state data) because they remove from the BOLD signal the main effects of your tasks, and that avoids confounding what could be "true" connectivity between two regions with "artifactual" connectivity resulting from coactivation of these two regions in response to your tasks/conditions.
Hope this helps
Alfonso
Originally posted by shounak dasgupta:
Yes, that is perfectly fine. If I am interpreting correctly that is the approach described in Fair, Damien A., et al. "A method for using blocked and event-related fMRI data to study "resting state" functional connectivity." Neuroimage 35.1 (2007): 396-405.
In general, when you have a task-related design and you are not interested in obtaining condition-specific connectivity measures, you should still enter the design information in Setup.Conditions and then simply right-click on your task conditions and select the option named "move selected condition to covariate list (Fair et al. resting state analyses of task-related data)". That will remove these task conditions from your conditions list but it will also create a set of new first-level covariates that describe the task-related main effects on the BOLD signal. When you get to the Denoising step, simply enter the "Effect of #" effects into the list of confounding effects in order to regress out these effects before computing any connectivity measures. After doing these steps you can continue with first-level analyses as you would do normally for a true "resting state" dataset, just analyzing the "rest"-condition connectivity. These steps are considered better than simply disregarding the condition information (and treating your task-related data as if it was a resting-state data) because they remove from the BOLD signal the main effects of your tasks, and that avoids confounding what could be "true" connectivity between two regions with "artifactual" connectivity resulting from coactivation of these two regions in response to your tasks/conditions.
Hope this helps
Alfonso
Originally posted by shounak dasgupta:
I have a block design paradigm with a rest block
between two stimulation blocks. I don't want to see between
conditions connectivity. I want to find a correlation map for
a seed's total time series ( without considering different
condition specific time series) . But I was thinking of approaching
the problem in a different way. I am taking a single time series of
the seed region ( both rest and activation) instead of condition
specific time series and have correlated it with similar time
series of every other voxel ( by defining a single condition at the
set up stage where onset= 0 and duration = inf).
Is this approach reasonable to follow? Or should I use condition specific correlation maps?
If this is fine then -
what will I select in weighting section of 1st level analysis?
Is this approach reasonable to follow? Or should I use condition specific correlation maps?
If this is fine then -
what will I select in weighting section of 1st level analysis?
Threaded View
| Title | Author | Date |
|---|---|---|
| shounak dasgupta | Jul 2, 2015 | |
| Alfonso Nieto-Castanon | Jul 4, 2015 | |