Hi Rick,
I found the following article that seems to indicate what you suggest:
[
www.ncbi.nlm.nih.gov]
I assume 3dDeconvolve implements an adaptive two-stage FDR approach as it is described in 3dFDR. According to the paper, would this mean that a two-stage FDR approach might not be appropriate if statistical values are very high in the entire brain?
Dataset was acquired at 7T and preprocessed as follows:
- RETROICOR, RVT, CR, and Legendre polynomials up to 4-th order
- Smoothing with 4mm FWHM Gaussian kernel
- Computation of signal percentage with 3dcalc -expr 'a - mean(a)/mean(a)'
Yes, please let me know how to upload the data.