I am trying to set up a regression using a combination of stimtimes_AM2 (label1=modulated hits) and stimtimes (label2=guess), with modeling using a tent function (0, 16, 9). The TR is 2 seconds. I would like to confirm how to interpret the output correctly.
The regression through 3dDeconvolve yields 18 coefficient-tstat files for label1, and 9 coefficient-tstat files for label2.
1.Is the output coefficient paired with output coefficient as the weighted and unweighted coefficients respectively (such that  goes with ,  goes with  etc), or are coefficients- the unweighted label1 outputs and coeffcients- weighted label1 outputs?
2.Given this output, what is the correct way to set up the glts?
I would like to set up one glt of the unweighted hits (label1) - guess (label2). Would the correct syntax be: ‘SYM: hits - guess’ or should I specify all coefficient files: ‘SYM: hits[0:8] -guess’?
Similarly, I would like to set up one glt of the weighted hits (label1) alone. Would the correct syntax be: ‘SYM: hits’ or ‘SYM: hits[9:17]’?
- Finally, how does stimtimes_AM2 contribute to the iresp files output. If I specify (-iresp 1 hits, for example), does the output represent the unweighted portion of label1, the weighted portion of label1, or the overall mean?
Thanks, I really appreciate the help.
I believe that the first 9 coefficients - are the estimated hemodynamic response at the average value of the modulation variable (average hits in your case), while the second set - are the modulation effects.
If you want to compare the two conditions in terms of the whole hemodynamic response curve, do this:
‘SYM: hits[[0…8]] -guess[[0…8]]’
Check out the details here: https://afni.nimh.nih.gov/pub/dist/doc/misc/Decon/DeconSummer2004.html
- I’m not so sure how the option -iresp works in this case, but you can simply extract those 9 coefficient using 3dbucket.
I tried setting up the gltsym with the double brackets as you suggested. ‘SYM: hits[[0…8]] -guess[[0-8]’. When I run 3dDeconvolve, the output now has 9 GLT files.
Is there a syntax where I get the sum or average of all the tent coefficient files so that the comparison of the whole hemodynamic responsive curve gives me one GLT coefficient file? I was expecting one GLT coefficient file, so that I could set up 3dMEMA like analyses.
‘SYM: hits[[0…8]] -guess[[0-8]’. When I run 3dDeconvolve, the output now has 9 GLT files.
Each of the nine tests shows the contrast at a particular time point. In addition, you should see an F-stat that indicates the overall or omnibus contrast between the two curves.
Is there a syntax where I get the sum or average of all the tent coefficient files so that the comparison of the
whole hemodynamic responsive curve gives me one GLT coefficient file?
For that, you can do
‘SYM: hits -guess’
‘SYM: 0.125hits -0.125guess’
I was expecting one GLT coefficient file, so that I could set up 3dMEMA like analyses.
That is not necessarily a good approach. Check out this: http://dx.doi.org/10.3389/fnins.2015.00375