help
488 Subscribers 
help > Single 2nd Level ROI-ROI Analysis with Some Missing ROI's
Showing 1-3 of 3 posts
Jan 31, 2018 09:01 PM | Benson Stevens
Single 2nd Level ROI-ROI Analysis with Some Missing ROI's
Hello,
I am working with functionally defined ROI's at the individual subject level using 4 different functional localizer scans. As you probably know, when doing this, sometimes an individual won't activate in a given region (say the visual word form area), thereby not creating a function ROI, but will in all the other ROI's. I am wondering if there is a way to still do a single 2nd level ROI-ROI analysis including those subjects? Someone told me I could enter "dummy" ROI's when setting up the project, then at the 2nd level input some vector with zeros for those subjects, so they are not included at all in a given analysis. However, that is not what I'm wanting. Here is a very simple example:
Say you have 3 subjects and creating 3 functionally defined ROI's for each (say maybe VWFA, TPJ and IFG).
You are able to define all 3 ROI's for subjects 1 and 2. However, for subject 3 only TPJ and IFG work.
Is there a way in a single network analysis of all three ROI's to get the connectivity between the TPJ and IFG including all 3 subjects, but for say the VWFA to TPJ, have it omit subject 3 and calculate it using only the first 2 subjects? Would this even be statically valid?
Or if I want to look at all three ROI's do I have to omit subject 3 since they are missing one of them?
I kindly thank you in advance,
Benson Stevens
I am working with functionally defined ROI's at the individual subject level using 4 different functional localizer scans. As you probably know, when doing this, sometimes an individual won't activate in a given region (say the visual word form area), thereby not creating a function ROI, but will in all the other ROI's. I am wondering if there is a way to still do a single 2nd level ROI-ROI analysis including those subjects? Someone told me I could enter "dummy" ROI's when setting up the project, then at the 2nd level input some vector with zeros for those subjects, so they are not included at all in a given analysis. However, that is not what I'm wanting. Here is a very simple example:
Say you have 3 subjects and creating 3 functionally defined ROI's for each (say maybe VWFA, TPJ and IFG).
You are able to define all 3 ROI's for subjects 1 and 2. However, for subject 3 only TPJ and IFG work.
Is there a way in a single network analysis of all three ROI's to get the connectivity between the TPJ and IFG including all 3 subjects, but for say the VWFA to TPJ, have it omit subject 3 and calculate it using only the first 2 subjects? Would this even be statically valid?
Or if I want to look at all three ROI's do I have to omit subject 3 since they are missing one of them?
I kindly thank you in advance,
Benson Stevens
Feb 1, 2018 01:02 PM | Stephen L. - Coma Science Group, GIGA-Consciousness, Hospital & University of Liege
RE: Single 2nd Level ROI-ROI Analysis with Some Missing ROI's
Dear Benson,
If I understand correctly your question, I think this is possible but only bidirectionally: you must exclude from both source and target of correlation the 3rd subject, as I am not sure it is possible to just exclude from target but not source.
To do that, you can create a 2nd-level covariate where your 2 subjects have 1 assigned and the 3rd one NaN (eg, [1 1 NaN]). Indeed, to exclude a subject, it must be set at NaN, as using 0 is not enough and will still account for the subject's variability during the correlation analysis (0 means that the subject is regressed out, but not excluded, NaN must be used for that purpose).
You can also double-check your design in CONN, in the lower-right corner there is the number of subjects (n=XX), when using the NaN covariate it should be n=2 in your case.
Hope this helps,
Stephen
If I understand correctly your question, I think this is possible but only bidirectionally: you must exclude from both source and target of correlation the 3rd subject, as I am not sure it is possible to just exclude from target but not source.
To do that, you can create a 2nd-level covariate where your 2 subjects have 1 assigned and the 3rd one NaN (eg, [1 1 NaN]). Indeed, to exclude a subject, it must be set at NaN, as using 0 is not enough and will still account for the subject's variability during the correlation analysis (0 means that the subject is regressed out, but not excluded, NaN must be used for that purpose).
You can also double-check your design in CONN, in the lower-right corner there is the number of subjects (n=XX), when using the NaN covariate it should be n=2 in your case.
Hope this helps,
Stephen
Feb 14, 2018 12:02 AM | Benson Stevens
RE: Single 2nd Level ROI-ROI Analysis with Some Missing ROI's
Hi Stephen,
Sorry for the slow reply. Thanx for the heads up about using NaN instead of zeros, makes total sense. I guess it sounds like I can't exclude subjects in a single analysis from some of the target to sources but keep them in other targets to sources. I would be a nice function. Say having 3 ROI's, and including all 15 subjects between two of them, but use 13 to the third one. Hope all is well.
Many Blessings,
Benson
Sorry for the slow reply. Thanx for the heads up about using NaN instead of zeros, makes total sense. I guess it sounds like I can't exclude subjects in a single analysis from some of the target to sources but keep them in other targets to sources. I would be a nice function. Say having 3 ROI's, and including all 15 subjects between two of them, but use 13 to the third one. Hope all is well.
Many Blessings,
Benson