help > RE: Thresholding on Conn vs Spm
Nov 24, 2022  04:11 PM | Alfonso Nieto-Castanon - Boston University
RE: Thresholding on Conn vs Spm
Dear Daniela,

If you are using in CONN the default thresholding options (two-tailed stats, voxel-wise p<.001, cluster-level p-FDR<.05), to get the same results in SPM you would need to:

1. define in SPM a new F- contrast (instead of a T- contrast) with the same contrast values as your original T-contrast; this is to have SPM focus on two-sided results, otherwise for T-contrasts SPM only focuses on the positive side of the contrast

and 2. when displaying the results from SPM select the option "uncorrected", then enter 0.001, and then in SPM "cluster-size" threshold enter the value indicated in CONN top-right corner under "k>###"; or alternatively just enter 0 and then simply disregard any clusters where the cluster-level p-FDR stats are below 0.05. 

Hope this helps
Alfonso

 
Originally posted by Daniela Rodriguez:
Dear Alfonso,

I am having trouble reporting the results I have acquired with conn.


1. When I export the results to SPM via an SPM.mat file, even when applying the same threshold on SPM (voxelwise: p-FDR 0.05 and cluster-size p-FWE 0.05), it shows me different clusters as Conn. This is additionally difficult when I try to load in the SPM.mat file into the Anatomy toolbox. Have you been able to work around that?

2. Due to my first problem I decided to evaluate my results on conn directly. When it comes to describing my cluster location for my paper, I can only find T-values for the maximum voxel. I cannot find information on local maxima voxels or their T-values, which is important for big clusters. Where on Conn can I find this information?


I thank you in advance for your assistance.

Best Regards,
Daniela Rodriguez M.

Threaded View

TitleAuthorDate
Daniela Rodriguez Nov 24, 2022
RE: Thresholding on Conn vs Spm
Alfonso Nieto-Castanon Nov 24, 2022
Daniela Rodriguez Nov 25, 2022