open-discussion > single subject analysis
Showing 1-3 of 3 posts
Display:
Results per page:
Aug 22, 2019  06:08 AM | julie Bernard
single subject analysis
Dear,

I want evaluate the differente connectivity pre and post treatment for a single subject. This patient had catatonia which that disappeared with Zolpidem. I have no anatomical a priori.
Regarding the second level analysis, i will use between-conditions contrast vector to [-1 1] and between-subject contrast vector to [1], that's it?

But, for the first level analysis, i don't know if it's necessary to use ICC, global correlation or MVPA? in my case, i don't know what is relevant...

thank you in advance.

Julie
Aug 23, 2019  06:08 AM | Jeff Browndyke
RE: single subject analysis
Hi, Julie.

With the current CONN version it is not possible to run single-subject designs, though here are two scripts that will allow for first-level comparisons between conditions.  In this case, you would establish a CONN analysis with one subject, two sessions (pre/post) and then run a (-1 1) comparison at the first level.  The output from the scripts are seed-to-voxel, so you will need to have some a priori atlas ROIs in mind.  I am hoping that someone has a solution for first-level, single-subject output that generates ROI-to-ROI maps, but I haven't seen any.  

You will not be able to get to any second-level designs in CONN with a single-subject...unless you trick it and enter the same subject twice.  Please bear in mind the lack of variance and significantly deficient model accomplished by doing this should not be used for reporting any results, but it may be helpful for rough pilot visualization of subject changes from pre [1 0] to post [0 1].

Hope this helps and also hope others may have single-subject type analysis ideas with CONN.

Warm regards,
Jeff
Sep 4, 2019  01:09 PM | julie Bernard
RE: single subject analysis
thank you for your response

Julie