Quantitive way to look if the model fits to the data

Dear AFNI experts,

I applied 3dDeconvolve to perform a linear regression at individual level between the tfMRI time series and the stimuli timing files from a task. The task is an event related design in which the subject looked at faces and scenes (randomly) during the run.
The results looks fine, higher peaks are located in fusiform areas for faces and in parahippocampal gyrus for scenes(image attached; faces are the above image and scenes are at the bottom). When I try to look to the model fit to the data in these areas the graph also looks coherent according what we are looking for.
Is there any way to know if model fits to the data in a quantitative way (calculating an specific value) instead of manually looking in each subject?

Thanks a lot,
Karel

Hi-

I think the full F-stat provides that information; among other things, it can be thought of as the ratio of the “fit” variance to the “unfit” variance.

That is why in the automatically generated afni_proc.py QC HTML output-- see here for a tutorial:
https://afni.nimh.nih.gov/pub/dist/doc/htmldoc/tutorials/apqc_html/main_toc.html
– the full F-stat is one of the outputs under the “vstat” (volumetric stat QC block) section.

–pt