reduction of ideals in 3dfim+

AFNI Message Board
1

Hello AFNI users,
I’m running a correlation analysis using the 3dfim+ command and I have some questions about the ideal function.
I’m currently using the whole amygdala as seed region, resulting in 172 ideals. I read in the 3dfim+ documentation that, when there are multiple ideals, “the kth ideal is referred to as the ‘Best Index’, and the program reports the parameters alpha, p, etc., for the Best Index only”.

I’d rather prefer to use as ideal an estimate of the time course of the seed region instead of 172 ideals. Is there a way to estimate the time course of the entire seed region (or a subset of voxels) and use then just one ideal as input in 3dfim+? would the result of this correlation be appropriate?

Thanks,
ns

2

While it will probably do the job, 3dfim+ is a very old program that we don’t usually recommend. Consider using 3dTcorrD for a correlation analysis together with 3dmaskave to get an average time course for a region.

3

Thanks, I’ll follow your suggestion then
ns

4

Daniel, as you recommended, I used 3dmaskave and 3dTcorr1D for the correlation analysis.

However, I still have a question about 3dfim+. Is there a way to get out the info about the best index (best kth ideal)? I’m wondering if I can get out a txt file containing more info about it.

thanks
ns

5

3dTcorr1D doesn’t have that functionality, but you could script it, probably with 3dTcat for the individual outputs, then 3dTstat -argabsmax (or similar output option) to find the dataset that gives the largest absolute correlation.