Dear AFNI experts,

My overall issue is that after running afni_proc, the registration between the T1 and the EPI images does not align well. Some chunks of the brain are not captured. Also for some subjects the EPI image is shifted upwards outside of the T1 space (pictures attached).

The functional data is oblique. Originally I ran afni_proc without addressing this and the registration looked worse than when I deobliqued it. I tried deoblique-ing the raw data before running afni_proc and also adding in -align_ops_aea -deoblique on. The registration looked better when I deobliqued it before running afni_proc than when I added it to the align options. Is it alright to do deoblique before afni_proc or will it affect the data processing?

Do you have any other recommendations on ways to improve the registration? I used align_epi_anat.py to try the different combinations of cost functions/moves. Below is the afni_proc call I am using.

afni_proc.py -subj_id ${subject} \
-out_dir ${subject}.deob \
-script proc.${subject} -scr_overwrite \
-blocks despike tshift align volreg \
-align_opts_aea -giant_move -cost lpc+zz \
-anat_has_skull no \ ##### images were skull-stripped using FSL BET
-despike_opts_3dDes -ignore 2 \
-copy_anat Anat/bet_anat.nii.gz \
-dsets deob_run?+orig.HEAD \
-tcat_remove_first_trs 2 \
-volreg_align_to MIN_OUTLIER \
-volreg_align_e2a \


Thank you for your time,
Cate