large file size since volreg in afni_proc

Dear afni experts:
I am using afni_proc to generate preprocessing script and perform preprocessing. I list afni_proc command as below. -subj_id sub-04
-script orientdecode.csh -scr_overwrite
-out_dir afniprep
-blocks despike tshift align volreg mask scale regress
-copy_anat forrestgump_sub-04_SurfVol.nii
-dsets sub-04_ses-r20_task-orientation_rec-dico_run-**_bold.nii.gz
-tshift_opts_ts -verbose -tpattern @SliceTiming.1D
-volreg_align_to MIN_OUTLIER
-regress_errts_prefix motpolyerrts
-regress_fitts_prefix motpolyfitts
-regress_opts_3dD -rout

My original nifti file for one function run is ~ 70mb. I see that file size (brik files) increases to ~180mb in tcat, despike and tshift steps. It is still OK. But file size increases to ~1GB since the volreg step. I have 10 runs and data occupy ~10GB space. Did I do something wrong? Is it normal? Is it possible to reduce the file size?



I might find the problem. It might due to a bad alignment between anat and vr_base_min_outlier+orig file. And the grid of functional data after volreg is much bigger than the original gird.

I’ll manually improve the alignment and see whether the size is still big.

The change to tcat might just be the compression. You could set


to get a similar effect. It could also grow if it shifts from shorts to floats for some reason, which might depend on the data.

The large scale after volreg is stranger. Are the dimensions far from isotropic? What is the output from

3dinfo -d3 -n4 sub-04_ses-r20_task-orientation_rec-dico_run-**_bold.nii.gz

and the same for pbvolreg.HEAD

  • rick

Thanks Rick!

Yes, it seems that at the tcat step, it goes from short to float That’s why it increases from ~80mb to 180mb. But it is OK.

The grid of data after volreg seems to expand a lot. See below

3dinfo -d3 -n4 sub-04_ses-r20_task-orientation_rec-dico_run-01_bold.nii.gz
2.000000 -2.000000 2.200000 100 100 37 121

3dinfo -d3 -n4 pb**volreg*.HEAD
2.000000 -2.000000 -2.000000 128 128 128 121

Any suggestion here?

Thanks again.