Different t-values found for comparison done in Rest vs SPM

Submitted by Kyle Heidtman on Mon, 07/01/2013 - 21:22

Hello,

I recently generated some correlation maps.  I then performed the Fisher Z transform on those maps.  I fed those z-maps into the Rest Statistical Analysis tool to perform t-tests to look for differences between the groups.

I also happened to input those z-maps into a 2nd level SPM analysis and did a t-test between the groups in that program. 

I compared the t-value maps generated by those two programs (I compared the T.nii file created by REST and the spmT_0001.hdr file created by SPM), and I found there were differences between the t-maps.  The differences were fairly small.  The mean difference is

Submitted by YAN Chao-Gan on Tue, 07/02/2013 - 00:03

Hi,

1. The output of REST is consistent with SPSS. The subttle difference between REST and SPM may caused by the data precision? I am not totally sure.

2. I don't understand cluster-based FDR (you may find it in SPM), I do understand cluster-based Gaussian Random Field Theory Correction. If you mean the latter, you can find a module in REST.

Best,

Chao-Gan


Submitted by YAN Chao-Gan on Tue, 07/02/2013 - 00:03

Hi,

1. The output of REST is consistent with SPSS. The subttle difference between REST and SPM may caused by the data precision? I am not totally sure.

2. I don't understand cluster-based FDR (you may find it in SPM), I do understand cluster-based Gaussian Random Field Theory Correction. If you mean the latter, you can find a module in REST.

Best,

Chao-Gan