Hi,
I am running group analyses on PET data. I’m finding that when I threshold at an uncorrected p-value of .0005, I get nice activations in several brain regions. However, absolutely nothing survives FDR or family-wise error correction (both set at .05, corrected). I believe the trouble lies with the fact that I’m not masking my images properly. I came to this conclusion by masking the images first in a different program and then bringing the data back into SPM. At this point, because the data are already masked, I request no further masking in SPM. After doing all this, I do get significant activity at p < .05, corrected.
I have tried several different masking procedures (an explicit mask using a structural T1 template, a threshold mask with a value of .80 of the mean) but nothing seems to work. Any ideas what I’m doing wrong? Is there an matlab script I might be able to look at?
Preprocessing steps:
1. Convert 40s summed image into one nifti file; two such files for each injection/condition for each subject. 30 subjects.
2. Warp to PET template
3. Smooth with 12mm filter
4. Paired t-test
Thanks for your help.
Jatin