Dear SPM users!
I have a few questions regarding viewing and interpreting the results in VBM.
My study consists of 20x20 subjects and I do a simple t-test to find differences in the amount of gray metter between the two groups and some correlation statistics with behavioral variables.
The questions are:
1) the results do not survive correction, nor FWE or FDR. I have seen a paper (Salgado-Pineda NeuroImage 2003) describing that vbm data are not robust enough to survive correction built in SPM (designed for functional data). Is it OK to view the results uncorrected? Did you encounter similar problem?
2) when I view the results with p<0,001 uncorrected I see only a few tiny blobs. With a less strict treshold the results are more meaningfull in the sense of literature data. Again there are some papers (Wright, Sch Research 1999) which consider significant results at the level of 0,01 uncorrected. So again, can I take the results as valid when viewing uncorrected p<0,01?
3) Is there another way to correct the results that are, letīs say, less strict?
Thank you for your reply!
Tomas Kasparek
Dep. of Psychiatry
Masaryk University
Brno, the Czech Republic
|