Dr. Phoebe Chan asks the following questions:
" I just read through your papers in J Cogn Neurosci 2003, Vol 15, p.991 and Neuron 2004, Vol 41, p.809. They are very nice papers. I'm also using AFNI to analyse the data in my language fMRI event-related study with two types of stimuli (words and tones). However, I got stuck in applying the two-step thresholding process that you mentioned in your papers on my data. Two questions on the single subject and group comparison analysis that I would like to seek your advice.
1) For the single subject analysis, after deconvolution using 3dDeconvolve, a full F-stat map, maps for individual stimulus and for the contrast of words vs. tones were generated. In the first step of thresholding process, I chose Full F-stat map for both 'Func' and 'Thr', and set the threshold to
p=1x10^-6 to cut off the voxels with large p values. After that, I chose the contrast 'words vs. tones' in the second step thresholding process to show the active voxels at p<0.04.
Is it the correct way to do the two-step thresholding process using AFNI? I saw some comments that stringent threshold was set for the F-map of individual stimulus in the first step of thresholding process instead. Do you think it's valid?
2) For the group comparison analysis, I did 3dANOVA for my set of data with the following options:
-mean 1 words \
-mean 2 tones \
-diff 1 2 words-tones \
-contr 1 1 All \
-contr 1 -1 wordsvstones \
-ftr StimEffect \
-bucket AvgANOVA_13subj
I'm not very sure whether I should apply the stringent threshold on the 'All' map generated with contrast of all the stimuli present, or on the 'StimEffect' map generated with the factor of all stimuli. The reason is that both of the maps accounted for all types of stimuli but showed a huge difference in terms of threshold values. Which map do you think it's more appropriate for the first step thresholding process and which one for the second step?"