How to view the positive and negative clusters

AFNI Message Board
1

How do I know which clusters are positive and which ones are negative when using ‘bi-sided’ option in Clusterize in AFNI GUI? I am only able to see the number of positive and negative clusters. If the number of the resulting clusters is 5, it only reports 2 positive and 3 negative but not which ones. I would also like to save +ve/-ve when saving the table from the report.

Any help is appreciated. I am sure there must be a simpler way than interpreting on the basis of color in the images or computing from Olay, Ulay and Thr after jumping to each cluster. Any option using the command line is also not a problem.

Thanks

2

Hi-

There is this column in the 3dClusterize tabular output:


Max Int      : Maximum Intensity value for the volume cluster

… so the sign of that should tell you which cluster is pos/neg when using bi-sided (generally recommended) clustering.

In the GUI, when you hit the “Save Mask” button in the Clusterize Report controller, you will see the 3dClusterize command run on the commandline for you, as well; you can redirect that table to the file.

Just in case you are bored and want some related reading on the topic of using stats/betas for different aspects of reporting, then you could take a look at:

  1. Is the statistic value all we should care about in neuroimaging?
    https://www.ncbi.nlm.nih.gov/pubmed/27729277
    (Hint about the title: the answer is “no”.) This describes why using more than stats is important, and the separate roles of the stats and betas in interpreting your results.

and

  1. A tail of two sides: Artificially doubled false positive rates in neuroimaging due to the sidedness choice with t-tests.
    https://www.ncbi.nlm.nih.gov/pubmed/30265768
    which is covers (among other useful bits) how bi-sided or two-sided testing can still allow you to tell which clusters are positive and negative.

and here might be a useful example of automatically taking snapshots of your stats results:
https://afni.nimh.nih.gov/pub/dist/doc/htmldoc/tutorials/auto_image/auto_@chauffeur_afni.html#ex-7-threshold-stats-voxelwise-clusterize-view-effects-ii

–pt