Hi, my EPI-to-anat registration QC output shows the ulay as more distorted than I would expect, given that I'm applying -blip correction in afni_proc. Multiband (7), single-echo EPI data were acquired with AP phase encoding; I have AP and PA fieldmaps that I'm using as forward_dset and reverse_dset respectively.
epi2anat QC images usually look ~like the below, across subjects in my dataset.
Are these results more or less the expected outcome (I know SDC can't fix everything), or is there a way to improve registration and mask accuracy? (or am I doing something incorrectly in the afni_proc pipeline....)
sub #1
sub #2
AFNI version info (24.1.22):
afni_proc.py \
-subj_id "${subject}_${session}_${task}" \
-dsets ${epi_run1} ${epi_run2} \
-copy_anat ${anat_cp} \
-scr_overwrite \
-anat_has_skull no \
-anat_follower anat_w_skull anat ${anat_skull} \
-blocks despike tshift align tlrc volreg mask blur scale regress \
-radial_correlate_blocks tcat volreg \
-blip_forward_dset ${fmap_ap} \
-blip_reverse_dset ${fmap_pa} \
-blip_opts_qw -plusminus -noXdis -minpatch 9 \
-tcat_remove_first_trs 0 \
-align_unifize_epi local \
-align_opts_aea \
-cost nmi \
-check_flip \
-tlrc_base MNI152_2009_template_SSW.nii.gz \
-tlrc_NL_warp \
-tlrc_NL_warped_dsets "${dsets_NL_warp[@]}" \
-volreg_align_to MIN_OUTLIER \
-volreg_align_e2a \
-volreg_tlrc_warp \
-volreg_warp_dxyz ${dxyz} \
-volreg_compute_tsnr yes \
-mask_import Tvent ${template_imp} \
-mask_segment_anat yes \
-mask_segment_erode yes \
-mask_intersect Svent Tvent CSFe \
-mask_epi_anat yes \
-blur_size ${blur_size} \
-blur_in_automask \
-regress_motion_per_run \
-regress_ROI_PC Svent 3 \
-regress_ROI_PC_per_run Svent \
-regress_make_corr_vols WMe Svent \
-regress_anaticor_fast \
-regress_censor_motion ${censor_mot} \
-regress_censor_outliers ${censor_out} \
-regress_apply_mot_types demean deriv \
-regress_est_blur_epits \
-regress_est_blur_errts \
-out_dir "${procDir}/${subject}/${session}/proc_sst" \
-html_review_style pythonic