3DClusterize gives different number of clusters

Dear all,

I have been trying to use the 3dClusterize tool on a simple Nifti file, but every time I run it I get a different amount of clusters.
I currently use the 3dClusterize command to get an output file with the clusters using different labels, and then use 3dcalc to separate each cluster into a different Nifti file. My question is, is there a reason why I am getting a specific amount of clusters? Is it related to the algorithm used to cluster?
Also, when setting threshold values with the -within_range flag, I get fewer clusters than, say, other available software. Do you might know why this could be happening?
Thanks in advance: looking forward to hearing from you!


Hi, Fernanda-

It would help to see the 3dClusterize commands you are using. There shouldn’t be anything stochastic/random in the algorithm to cause underlying variability. Are you certain that you are using the exact same threshold (you can rightclick the Thr above the colorbar, and select “Set threshold” or “Set p-value” directly by typing for example) and parameters (cluster size, NN value, and choice of bisided/2sided) each time?

  • You can output the 3dClusterize command that Clusterizing is using in the report, by clicking on the “3dclust” button in the top button row of the report. Can you copy+paste those here, so we can see if anything is different?

I am not aware what other software has something like the “-within_range …” functionality, actually. What are those other programs you are comparing with? Again, it would help to see both the AFNI commands and the other ones to be more sure.