Dear AFNI users,

I am currently still struggling to get correct data from my Man Whitney test. I had mentioned previously that I was having trouble extracting the correct voxel volumes from the test, and your advice was very helpful.

To recap:

1) Performed Mann Whitney test on 2 sets of patients

2) Ran an AlphaSim test on a mask with the desired dimensions using the following command:

AlphaSim -mask MASK+tlrc -fwhm 10 -rmm 1.42 -pthr 0.001 -iter 2000 -out AlphaSimoutfile

3) The results of the AlphaSimoutfile gave me the cluster volume for an alpha size of .05 and smaller. This volume is 1058 mm^3.

4) Attempted to use 3dMerge to take out volumes of less than 1058 mm^3. I had tried this command at first

3dmerge -1clust 1.42 1048 -prefix 3DMergeresult+tlrc MannWhitneytest+tlrc

After running 3dclust on the 3dMergeresult file, I realized I had only one large cluster of volume over 600,000.

One user suggested I specify the -1thresh option to "threshold data to censor the intensities" as described in the user manual. Trying several threshold values, I produced bicks with several different cluster volumes.

The command I most recently used was:

3dmerge -1clust 1.42 1058 -1thresh .25 -1noneg -1dindex 1 -1tindex 1 -prefix 3DMergeresult_new+tlrc MannWhitneytest+tlrc

so that I would only be accessing the second subrick of the MannWhitneytest, using non-negative voxels, and isolating volumes of >= 1058


the Problem:

Now I am just not sure which threshold value would be appropriate for the results I need. If I want to isolate the Z statistic of the Mann Whitney test, have a p-value of .05 or .001, and use volumes of >=1058, which threshold value should I specify?

Are there flaws in my thinking or procedure so far? If any of the steps do not make sense please let me know.

Thank you for your help.

Regards,
Ellen Shlossberg