Template-space bold mask clipping in child data

[sgiavasis]

• The space-template_desc-bold_mask produced by single-step resampling has some missing sections on the bottom, like it's clipped.
• The native/bold-space bold mask looks good when overlaid over the sbref.

[tergeorge]

1. Looked at the various skull stripping strategies used to see if its causing the issue. BET skull stripping is broken and we need to add the robustfov before the BET step. This is not the main reason for the clipping of the mask because even though 3dskull stripping works we even see this issue when 3dskullstrip is used.
2. Looked at single step resampling. The antsApplyTransforms for the brain mask did not use the motion and the blip transform which is used for the functional timeseries. Checked if adding the two transforms to the brain mask would make a difference but the output did not change.

The gray regions in the image show the overlapping regions on the brain mask with and without the addition of the 2 transforms (motion correction and identity). As you can see from the image, the outputs do not change.

[sgiavasis]

@tergeorge currently also running fmriprep on this data:

• fmriprep-LTS without Freesurfer
• fmriprep 20.0.7

[tergeorge]

Current status: Ran fmriprep-LTS and fmriprep 20.0.7 on lisa. Run command: /data3/cnl/mindboggle/opt/singularity/bin/singularity run -B /data3/cnl/tgeorge/Leonas_issue:/data3/cnl/tgeorge/Leonas_issue -B /data3/aki/Leona/cpac_output/sample_data_and_fmriprepoutput/BIDs-raw:/data3/aki/Leona/cpac_output/sample_data_and_fmriprepoutput/BIDs-raw -B /home/agutierrez/Documents/C-PAC/dev/docker_data --cleanenv /data3/cnl/tgeorge/Leonas_issue/images/fmriprep-20.2.7.sif /data3/aki/Leona/cpac_output/sample_data_and_fmriprepoutput/BIDs-raw /data3/cnl/tgeorge/Leonas_issue/output/fmriprep_LTS_sub-PA009 participant --participant-label sub-PA009 -t ovalfamily --skip-bids-validation --fd-spike-threshold 0.9 --dvars-spike-threshold 2.0 --bold2t1w-dof 12 --output-spaces MNI152NLin6Asym:res-2 MNI152NLin2009cAsym:res-2 --stop-on-first-crash --notrack -w /data3/cnl/tgeorge/Leonas_issue/work/fmriprep_LTS_sub-PA009 --skip_bids_validation --fs-license-file /home/agutierrez/Documents/C-PAC/dev/docker_data/license.txt --fs-no-reconall --nthreads 8 --omp-nthreads 8 --mem-mb 12000

Singularity image error: Error message: /data3/cnl/mindboggle/opt/singularity/bin/singularity run -B /data3/cnl/tgeorge/Leonas_issue:/data3/cnl/tgeorge/Leonas_issue -B /data3/aki/Leona/cpac_output/sample_data_and_fmriprepoutput/BIDs-raw:/data3/aki/Leona/cpac_output/sample_data_and_fmriprepoutput/BIDs-raw -B /home/agutierrez/Documents/C-PAC/dev/docker_data --cleanenv /data3/cnl/tgeorge/Leonas_issue/images/fmriprep-20.2.7.sif /data3/aki/Leona/cpac_output/sample_data_and_fmriprepoutput/BIDs-raw /data3/cnl/tgeorge/Leonas_issue/output/fmriprep_LTS_sub-PA009 participant --participant-label sub-PA009 -t ovalfamily --skip-bids-validation --fd-spike-threshold 0.9 --dvars-spike-threshold 2.0 --bold2t1w-dof 12 --output-spaces MNI152NLin6Asym:res-2 MNI152NLin2009cAsym:res-2 --stop-on-first-crash --notrack -w /data3/cnl/tgeorge/Leonas_issue/work/fmriprep_LTS_sub-PA009 --skip_bids_validation --fs-license-file /home/agutierrez/Documents/C-PAC/dev/docker_data/license.txt --fs-no-reconall --nthreads 8 --omp-nthreads 8 --mem-mb 12000

Next steps: Build the image on ACCESS and run fmriprep again with the new singularity images.

[tergeorge]

ACCESS run command: LTS run: singularity run --cleanenv /ocean/projects/med220004p/tgeorge/Leonas_issue/images/fmriprep_20.2.7.sif /ocean/projects/med220004p/tgeorge/Leonas_issue/input/BIDs-raw /ocean/projects/med220004p/tgeorge/Leonas_issue/output/fmriprep_LTS_sub-PA009 participant --participant-label sub-PA009 -t ovalfamily --skip-bids-validation --fd-spike-threshold 0.9 --dvars-spike-threshold 2.0 --bold2t1w-dof 12 --output-spaces MNI152NLin6Asym:res-2 MNI152NLin2009cAsym:res-2 --stop-on-first-crash --notrack -w /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_LTS_sub-PA009 --skip_bids_validation --fs-license-file /ocean/projects/med220004p/tgeorge/Leonas_issue/images/fs_license.txt --fs-no-reconall --nthreads 8 --omp-nthreads 8 --mem-mb 12000

fmrprep: 20.0.7

singularity run --cleanenv /ocean/projects/med220004p/tgeorge/Leonas_issue/images/fmriprep_20.0.7.sif /ocean/projects/med220004p/tgeorge/Leonas_issue/input/BIDs-raw /ocean/projects/med220004p/tgeorge/Leonas_issue/output/fmriprep_LTS_sub-PA009 participant --participant-label sub-PA009 -t ovalfamily --skip-bids-validation --fd-spike-threshold 0.9 --dvars-spike-threshold 2.0 --bold2t1w-dof 12 --output-spaces MNI152NLin6Asym:res-2 MNI152NLin2009cAsym:res-2 --stop-on-first-crash --notrack -w /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA009 --skip_bids_validation --fs-license-file /ocean/projects/med220004p/tgeorge/Leonas_issue/images/fs_license.txt --fs-no-reconall --nthreads 8 --omp-nthreads 8 --mem-mb 12000

Output dir of runs: /ocean/projects/med220004p/tgeorge/Leonas_issue/output Working dir of runs: /ocean/projects/med220004p/tgeorge/Leonas_issue/work/

[tergeorge]

Fmriprep work directory outputs: https://docs.google.com/document/d/1k0O-IH-5ZmXM2EHhKg4v_BFatw4kKRduUQc5dPClGms/edit

[tergeorge]

Template space bold mask generation in fmriprep:

antsApplyTransforms --default-value 0 --float 1 --input /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/sdc_estimate_wf/pepolar_unwarp_wf/enhance_and_skullstrip_bold_wf/combine_masks/ref_bold_corrected_trans_corrected_brain_mask_maths.nii.gz --interpolation MultiLabel --output ref_bold_corrected_trans_corrected_brain_mask_maths_trans.nii.gz --reference-image /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_t1_trans_wf/gen_ref/sub-PA048_ses-V3W3_acq-MPR_rec-vNavNorm_T1w_corrected_masked_xform_reference.nii.gz --transform /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_reg_wf/fsl_bbr_wf/fsl2itk_fwd/affine.txt

Template space bold mask generation in CPAC:

antsApplyTransforms --default-value 0 --float 0 --input /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/_scan_ovalfamily/combine_masks_112/ref_bold_corrected_brain_mask_maths.nii.gz --interpolation MultiLabel --output ref_bold_corrected_brain_mask_maths_trans.nii.gz --reference-image /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/resampled_T1w-brain-template-funcreg/T1w-brain-template-funcreg/tpl-MNI152NLin2009cAsym_res-02_T1w_reference_resample.nii.gz --transform /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/ANTS_T1_to_template_58/write_composite_xfm/from-T1w_to-template_mode-image_xfm.nii.gz --transform /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/_scan_ovalfamily/convert_bbr2itk_200/affine.txt

[sgiavasis]

Splitting these up:

fmriprep:

antsApplyTransforms

--default-value 0

--float 1

--input /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/sdc_estimate_wf/pepolar_unwarp_wf/enhance_and_skullstrip_bold_wf/combine_masks/ref_bold_corrected_trans_corrected_brain_mask_maths.nii.gz

--interpolation MultiLabel

--output ref_bold_corrected_trans_corrected_brain_mask_maths_trans.nii.gz

--reference-image /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_t1_trans_wf/gen_ref/sub-PA048_ses-V3W3_acq-MPR_rec-vNavNorm_T1w_corrected_masked_xform_reference.nii.gz

--transform /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_reg_wf/fsl_bbr_wf/fsl2itk_fwd/affine.txt

CPAC:

antsApplyTransforms

--default-value 0

--float 0

--input /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/_scan_ovalfamily/combine_masks_112/ref_bold_corrected_brain_mask_maths.nii.gz

--interpolation MultiLabel

--output ref_bold_corrected_brain_mask_maths_trans.nii.gz

--reference-image /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/resampled_T1w-brain-template-funcreg/T1w-brain-template-funcreg/tpl-MNI152NLin2009cAsym_res-02_T1w_reference_resample.nii.gz

--transform /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/ANTS_T1_to_template_58/write_composite_xfm/from-T1w_to-template_mode-image_xfm.nii.gz

--transform /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/_scan_ovalfamily/convert_bbr2itk_200/affine.txt

• The float input parameter is different (check what this does?)
• Well, clearly the usage of transforms is different - fmriprep seems to have the one converted coregistration xfm, CPAC is using that plus the T1w-to-template?

In the meantime-

Can we see an overlay of the two input files, plus a correlation of the two? /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/sdc_estimate_wf/pepolar_unwarp_wf/enhance_and_skullstrip_bold_wf/combine_masks/ref_bold_corrected_trans_corrected_brain_mask_maths.nii.gz vs. /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/_scan_ovalfamily/combine_masks_112/ref_bold_corrected_brain_mask_maths.nii.gz

And same with the two references? /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA048/fmriprep_wf/single_subject_PA048_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_t1_trans_wf/gen_ref/sub-PA048_ses-V3W3_acq-MPR_rec-vNavNorm_T1w_corrected_masked_xform_reference.nii.gz vs. /outputs/working/pipeline_default-fmriprep_fs-off_be-bet_seg-ants_despike-off_nightly/cpac_sub-PA048_ses-V3W3/resampled_T1w-brain-template-funcreg/T1w-brain-template-funcreg/tpl-MNI152NLin2009cAsym_res-02_T1w_reference_resample.nii.gz

[tergeorge]

This is what the two inputs look like:

The one on top with the multicolor border is from fmriprep. correlation of the input files: 0.945126

For the reference image, CPAC uses the whole head while fmriprep uses just the brain template. So I can't overlay the two. Also need to investigate why that is the case.

CPAC reference:

fmriprep reference:

[tergeorge]

These are the two outputs overlaid on top of each other. The green is the fmriprep output. As you can see both of them are pretty bad!!

The two outputs have a different dimensions. CPAC - [82x97x82] and fmriprep - [62x76x61].

[tergeorge]

--float 1 means that the output image should be stored as a floating point.

[tergeorge]

fmriprep:

antsApplyTransforms --default-value 0 --float 1 --input /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA009/fmriprep_wf/single_subject_PA009_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_bold_trans_wf/bold_reference_wf/enhance_and_skullstrip_bold_wf/combine_masks/ref_bold_corrected_brain_mask_maths.nii.gz --interpolation MultiLabel --output ref_bold_corrected_brain_mask_maths_trans.nii.gz --reference-image /jet/home/tgeorge/.cache/templateflow/tpl-MNI152NLin2009cAsym/tpl-MNI152NLin2009cAsym_res-02_T1w.nii.gz --transform /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA009/fmriprep_wf/single_subject_PA009_wf/anat_preproc_wf/anat_norm_wf/_template_MNI152NLin2009cAsym/registration/ants_t1_to_mniComposite.h5 --transform /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA009/fmriprep_wf/single_subject_PA009_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_reg_wf/fsl_bbr_wf/fsl2itk_fwd/affine.txt

Breaking up the command:

antsApplyTransforms

--default-value 0

--float 1

Input : /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA009/fmriprep_wf/single_subject_PA009_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_bold_trans_wf/bold_reference_wf/enhance_and_skullstrip_bold_wf/combine_masks/ref_bold_corrected_brain_mask_maths.nii.gz ref: /jet/home/tgeorge/.cache/templateflow/tpl-MNI152NLin2009cAsym/tpl-MNI152NLin2009cAsym_res-02_T1w.nii.gz

--interpolation MultiLabel

--transform /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA009/fmriprep_wf/single_subject_PA009_wf/anat_preproc_wf/anat_norm_wf/_template_MNI152NLin2009cAsym/registration/ants_t1_to_mniComposite.h5

--transform /ocean/projects/med220004p/tgeorge/Leonas_issue/work/fmriprep_2007_sub-PA009/fmriprep_wf/single_subject_PA009_wf/func_preproc_ses_V3W3_task_ovalfamily_wf/bold_reg_wf/fsl_bbr_wf/fsl2itk_fwd/affine.txt

[tergeorge]

The two inputs overlaid.

The one on top with the multicolor border is from fmriprep.

[tergeorge]

The two outputs overlaid:

The image with multicolor border is from fmriprep.

[tergeorge]

The two references:

[tergeorge]

Lucid chart flow diagram of the CPAC and fmriprep work flow. https://lucid.app/lucidchart/373d7542-1de0-4c12-bf84-7de997dd99a5/edit?invitationId=inv_277f6ed3-42da-4573-bb97-791bcb11cde6&page=0_0#