gltsym option without redoing deconvolution

AFNI Message Board
1

Greetings,

I am wondering if it is possible to use the gltsym option to calculate a contrast on the single subject level on an existing statistical BRIK output from a previous 3ddeconvolve operation. More specifically, I calculated single trial beta values using the stim_times_IM option, and I now want to calculate a linear trend across each subsequent trial of each condition. I know that I could redo this in a new 3ddeconvolve script with these gltsym options added, but this would require redoing all steps of the single trial beta calculation which is a bit time consuming over many participants. I also know I could code this in 3dcalc, but this option feels clunky compared to using the gltsym options. Any advice regarding 3ddeconvolve or performing this in 3dcalc is appreciated.

2

Adding GLTs to an existing result requires first running the program with -xsave, after which one can use -xrestore.

On the flip side, if you want to model the trend among the IM betas, that could be done by running a new regression on those betas.

  • rick
3

Thanks so much for the reply, I hadn’t noticed the xsave option before which is very useful.

For running new regression on betas, may that also be done using 3ddeconvolve? This is perhaps a silly question, but I am just wondering if there are any extra steps (extra option flags?) needed to run 3ddeconvolve on non-BOLD timeseries BRIK files. Or, may I simply input a statistical BRIK file, and specify how to weight each subbrik with “-stim_file”?

thanks again.