Hello, 

I have 2 quick questions: 

I understand that the recommended statistical threshold is Z>1 and p>.005; How can I set this Z threshold? My results automatically return Z > 2.56.

After thresholding, there are 2 files: uncorrp_tfce, and uncorrp_voxel. I believe the former is FWE corrected, but what is the latter? I can't find literature on the difference. When visualising, the activations of the corrp_voxel seem to be more restricted. Moreover, which is "best" to use? 

Best,
Molly

Hi Molly,
Originally posted by Molly Rowlands:

I'm not sure I follow your question. What operation are you trying to perform?



Files named uncorrp_ are uncorrected p-values. uncorrp_tfce is the uncorrected p-value map of the tfce-map. uncorrp_voxel is the uncorrected p-value of the z-map, voxel-wise.

Hope this helps, kindly,

Anton

This helps - thank you!

Dear Anton,

I have a similar question that needs your further explanation. While the "corrp_tfce" is default and recommended, it usually report only one or two very big blobs. If using stricter probability thresholds, no region would survived. For example, I have done a mean analysis, and when I set the probability as 0.0005 and 10 extent voxels, a single region with 54899 voxels survived, but when I set the probability as 0.0004, 10 extent voxels, no region survives. It looks problematic to report a single region that include almost the whole brain voxels. 

With another option "corrp_voxel", when I set the probability as 0.001 and 10 extent voxels, more peaks with smaller blobs are reported, which makes more sense to me. Could you please explain how this correction is achieved?

One more question is, how many extent voxels is acceptible? The default is 10, but can I use other values?

Thanks a lot!

Guochun Yang