Hi,
I’m having issues getting 3dttest++ to work properly for a seed based analysis.
We have scans and mask files which we then manipulate using 3dmaskave, 3dTcorr1D, and 3dcalc to get a z corrmap. After that I use 3dcalc to subtract the followup scan from the baseline scan. All of the image files look good up to this point.
We then run 3dttest++ on all the files with some covariates and specifying -Clustsim, but the resulting image just shows up as a black screen in afni. One thing potentially of note is I’m inputting Nifti files and getting back AFNI files, but other than that I’m having issues even trying to trouble shoot what the error would be. The command seems similar to ones previous people have run in my lab.
Any help would be greatly appreciated. Thanks.
Hi-
Can you please post your 3dttest++ command?
There should be nothing related to BRIK/HEAD vs NIFTI affecting things here. AFNI both reads and writes NIFTI, as well as BRIK/HEAD. If you would like the output dataset to be a NIFTI volume, you can just put .nii or .nii.gz at the end of your output prefix name, like:
-prefix NAME.nii.gz
That is true of most AFNI programs (and some have a -nifti option for output, typically when there are multiple output files).
–pt
I thought that was the case, but that was my only idea. I also double checked the diff files I created and they seemed to look correct (or at least similar to ones my lab created in the past).
3dttest++ -setA group \
subj# PATH/TO/FILE
…
-prefix group_diff_ttest_covar_clustsim_thalamusmask
-covariates tx_response.txt
-Clustsim
OK, thanks.
The output dataset will likely have multiple volumes—you can see the labels in the 3dinfo output:
3dinfo -label group_diff_ttest_covar_clustsim_thalamusmask
3dinfo -verb group_diff_ttest_covar_clustsim_thalamusmask
You can also check the min/max within the datasets:
3dinfo -min -max group_diff_ttest_covar_clustsim_thalamusmask
Are those really identically zero, or just small?
Also, with -Clustsim, are you using a mask, as well?
Note that the Clustsim aspect here doesn’t change the values of your t-test results—those should just be standard t-tests. What it does do is produce a series of *.1D files that contain the cluster-calculation to apply, e.g., with the Clusterize plugin in the GUI or with 3dClusterize on the commandline.
–pt
When I used 3dinfo -label, I had already attempted to view these results when I checked in the AFNI window to define the overlay and they appeared to be 0.
For 3dinfo -min -max, the values are identially 0. I didn’t specify a mask using clustsim. I did use a mask to get the z corr maps. I thought if I’m not specifying a mask it would just look at the whole brain and would essentially just be more data than strictly needed, but not incorrect.
I see the 1D files and when I launch afni from the terminal on the output file it gives me a message saying that it found Clustsim tables.
So, if you run 3dttest++ without -Clustsim, you also just get an empty set of data, is that right?
Can you check your mask, and make sure that it is where you think it is/should be?
–pt
Hi, I was able to get everything to work by resampling the image and the seeds to a template before running my process. Thanks for your help.