Hi folks,
I am trying to align fMRI data to the TT_icmb template. The alignment works out quite well, for the most part but there seems to be a slight misalignment in the frontal lobe towards the top. I'd be grateful, if the AFNI literates here could point to something in my script that could possibly be causing this? Also, what could be some other factors leading into this?
Thank you!!
And here's the script:
#######################################
# Spatially normalize the anatomical dataset to #
#######################################
# Uniformly distribute the white matter in the brain tissue.
3dUnifize -prefix BOLD-${SUBJECT_ID}-EC-T1 \
-input ANAT-${SUBJECT_ID}.nii
# Remove the skull and extract the brain tissue from T1-weighted MR
# image.
3dSkullStrip -prefix ${SUBJECT_ID}-T1-NoSkull \
-input BOLD-${SUBJECT_ID}-EC-T1+tlrc.
# Shift the center of DSET to the center of BASE. '_shft' will be
# appended at the end of DSET. Use the center of mass of the volume as
# the center (By default, center is the center of volume's grid).
@Align_Centers -cm \
-base ${SUBJECT_ID}-BOLD-EC-base+orig. \
-dset ${SUBJECT_ID}-T1-NoSkull+tlrc.
# Linearly align the anatomical dataset with the EPI dataset. The
# EPI dataset cab be the functional-MR image. The -epi_base option
# specifies the starting sub-brick for the alignment.
align_epi_anat.py -anat ${SUBJECT_ID}-T1-NoSkull_shft+tlrc. \
-epi ${SUBJECT_ID}-BOLD-EC-base+orig. \
-epi_base 0 \
-suffix _alBOLDEC \
-anat_has_skull no \
-epi_strip 3dAutomask \
-volreg off \
-tshift off \
-resample off
# Shift (roughly) the center of the aligned anatomical dataset with the
# standard template in the MNI space. The transformation information
# will be stored in the base image of BOLD data as well.
# Again, "_shft" will be appended to the DSET and CHILD.
# Create a symbolic link for the atlas you want to use.
# We will use TT_icbm452+tlrc. atals.
ln -sf /opt/afni/TT_icbm452+tlrc.* ./
@Align_Centers -base TT_icbm452+tlrc. \
-dset ${SUBJECT_ID}-T1-NoSkull_shft_alBOLDEC+tlrc. \
-child ${SUBJECT_ID}-BOLD-EC-base+orig.
# Align SOURCE to BASE and save the transformation matrix for each
# sub-brick into a 1D file. The cost function that defines the matching
# between SOURCE and BASE is the lpa (Local Pearson Correlation Abs).
# Also compute a weight function using 3dAutomask alogrithm plus some
# blurring of the base image.
3dAllineate -prefix ${SUBJECT_ID}-T1_to_T1_Allineate \
-base TT_icbm452+tlrc. \
-source ${SUBJECT_ID}-T1-NoSkull_shft_alBOLDEC_shft+tlrc. \
-1Dmatrix_save T1_to_T1_Allineate.aff12.1D \
-source_automask \
-cost lpa \
-autoweight \
-cmass
# Compute a non-linearly warped version of SOURCE dataset to match the
# BASE dataset. Gaussian blur the SOURCE and BASE before doing the
# alignmnet. By default, the blur values is 2.345 (for no good reason).
# Set the blur values to 0 if you do not want to blur the inputs.
3dQwarp -prefix T1-NoSkull_shft_alBOLDEC_shft \
-blur 0 0 \
-base TT_icbm452+tlrc. \
-source ${SUBJECT_ID}-T1_to_T1_Allineate+tlrc
# So, the 3dcopy command seems to overwrite one of the files created
# above. The old file needs to be deleted for 3dcopy to execute.
# ${SUBJECT_ID}-T1-NoSkull_shft_alBOLDEC_shft+tlrc.* will be recreated
# by 3dcopy.
rm --force ${SUBJECT_ID}-T1-NoSkull_shft_alBOLDEC_shft+tlrc.*
# Copy one dataset to the other. 3dCopy foo bar copies foo+orig. to
# bar+orig. or foo+tlrc. to bar+tlrc. This program copies entire
# datasets and not just sub-bricks.
3dcopy T1-NoSkull_shft_alBOLDEC_shft \
${SUBJECT_ID}-T1-NoSkull_shft_alBOLDEC_shft
##################################
# Spatially normalize the BOLD dataset #
##################################
# Create a copy of the original volume registartion mask.
3dcopy ${SUBJECT_ID}-BOLD-EC-volreg-mask+orig. \
${SUBJECT_ID}-BOLD-EC-volreg-mask_temp
# Shift (roughly) the center of the 4D BOLD fMRI data to the space
# using the same parameters as the base image. The base image has
# similar alignmnet as the anatomical image. (As explained earlier).
@Align_Centers -cm \
-base ${SUBJECT_ID}-BOLD-EC-base_shft+orig \
-dset ${SUBJECT_ID}-BOLD-EC-volreg-mask_temp+orig.
## Apply a nonlinear transformation and resample to 3x3x3 mm
3dNwarpApply -prefix ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp \
-source ${SUBJECT_ID}-BOLD-EC-volreg-mask_temp_shft+orig. \
-nwarp 'T1-NoSkull_shft_alBOLDEC_shft_WARP+tlrc. '
# 'T1_to_T1_Allineate.aff12.1D'
3dcopy ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp+orig \
${SUBJECT_ID}-BOLD-EC-volreg-Nwarp_Backup
3drefit -view tlrc ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp+orig
##get tissuse-based signal before smooth
3dDeconvolve -float -polort A \
-errts ${SUBJECT_ID}-BeforeSmooth-lp \
-bucket BeforeSmooth-bucket \
-input ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp+tlrc.
3dBandpass -band 0.01 0.08 \
-prefix ${SUBJECT_ID}-BeforeSmooth-lp-bp \
-input ${SUBJECT_ID}-BeforeSmooth-lp+tlrc
# Spatial smoothing with 6 mm FWHM. The value for FWHM is adjustable
3dmerge -doall \
-1blur_fwhm 6 \
-prefix ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp-blur6 \
${SUBJECT_ID}-BOLD-EC-volreg-Nwarp+tlrc
# Linear detrending
3dDeconvolve -float -polort A \
-bucket bucket \
-errts ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp-blur6-lp \
-input ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp-blur6+tlrc.
# temporal filtering band pass 0.01-0.1 Hz
# you can modify the filter if you you want
3dBandpass -band 0.01 0.08 \
-prefix ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp-blur6-lp-bp \
-input ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp-blur6-lp+tlrc
3dresample -master TT_icbm452+tlrc. \
-prefix ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp-blur6-lp-bp-resampled \
-input ${SUBJECT_ID}-BOLD-EC-volreg-Nwarp-blur6-lp-bp+tlrc.
