ASL Pipeline for the Human Connectome Project

This repository contains the ASL processing pipeline scripts for the Human Connectome Project.

Contents

• Prerequisites
• Installation

Prerequisites

The HCP list some prerequisites for their pipelines: https://github.com/Washington-University/HCPpipelines/wiki/Installation-and-Usage-Instructions.

The prerequisites specific to this pipeline, along with links to their installation pages, are listed below:

• FSL (version >= 6.0.5.1)
• Workbench (version >= 1.5.0)
• HCP Pipelines
• FreeSurfer

FSL version 6.0.5.1 is required to use new features in fabber and oxford_asl_roi_stats.py, which were added for this pipeline.

Workbench >= v1.5.0 is required. The -weighted option in wb_command's -volume-to-surface-mapping introduced in version 1.5.0 is used in projections to the surface. The environment variable CARET7DIR should point to directory containing the wb_command executable.

The HCP Pipelines must be installed and the environment variable HCPPIPEDIR set in order for the (Sub-)Cortical LUTs to be used in SE-based bias correction.

FreeSurfer must be installed to enable boundary-based registration (BBR). The environment variable FREESURFER_HOME must be set.

Installation

It is advised that the pipeline is installed in a conda environment. Python >= 3.9 is required. Note that some dependencies require cython and numpy to be installed prior to installing the hcp-asl package, for example following the steps below.

conda create -n hcpasl python=3.11 cython 'numpy>=1.25'
conda activate hcpasl
pip install git+https://github.com/physimals/hcp-asl.git 

Usage

Once installed, the pipeline may be run as a command-line script as follows:

process_hcp_asl --subid ${Subjectid} --subdir ${SubDir} --mbpcasl ${mbpcasl} --fmap_ap ${SEFM_AP} --fmap_pa ${SEFM_PA} --grads ${GradientCoeffs}

The filepaths passed to the script may be relative or absolute. A more detailed explanation of the arguments, including optionals, can be found by running:

process_hcp_asl --help

Custom perfusion quantification

To generate fully-corrected ASL data to use with other quantification tools besides oxford_asl, the pipeline can be run up to stage 8 as follows:

process_hcp_asl \
--subid ${SubjectID} \
--subdir /path/to/study/dir/${SubjectID} \
--grads /path/to/coeff_AS82_Prisma.grad \
--mbpcasl /path/to/mbPCASLhr_PA.nii.gz \
--fmap_ap /path/to/PCASLhr_SpinEchoFieldMap_AP.nii.gz \
--fmap_pa /path/to/PCASLhr_SpinEchoFieldMap_PA.nii.gz \
--stages 0 1 2 3 4 5 6 7 8 

The following outputs will be produced in /path/to/study/dir/${SubjectID}/T1w/ASL:

• asl_corr.nii.gz fully corrected ASL timeseries
• asl_corr_subtracted.nii.gz motion-and-banding-aware subtracted ASL timeseries
• TIs/timing_img_aslt1w.nii.gz effective TI (including slice-time) for ASL timeseries
• calib_corr.nii.gz fully corrected calibration image
• Calib/Calib0/calib_aslt1w_timing.nii.gz effective TI (including slice-time) for calibration image
• PVEs/ partial volume estimates for ASL timeseries

Pipeline stages

The pipeline can be run in stages via the --stages a b c argument. The stages are numbered 0 to 13 inclusive and perform the following operations. Note, each stage assumes the previous stages have run successfully, and will raise various errors if the outputs of those stages cannot be found.

0. Split mbPCASL sequence into ASL series and M0 images.
1. Derive gradient and susceptibility distortion correction
2. Correct M0 image
3. Correct ASL image
4. Label-control subtraction in ASL space
5. Perfusion estimation in ASL space
6. Registration and resampling of ASL/M0 into ASL-gridded T1w space
7. Partial volume estimation in ASL-gridded T1w space
8. Label-control subtraction in ASL-gridded T1w space
9. Perfusion estimation in ASL-gridded T1w space
10. Summary statistics within ROIs
11. Volume to surface projection
12. Copy key results into $outdir/T1w/ASL and $outdir/MNINonLinear/ASL
13. Create QC report

Extended pipeline outputs

Below is a directory tree within a subject's {SUBJID}_V1_MR folder showing the location of final and intermediate pipeline outputs. The default output directory has been used, so outputs are stored within the {SUBJID}_V1_MR folder.

SUBJECTID_V1_MR
ASL # Native acquisition space processing 
├── Calib # Intermediate calibration image processing 
│   ├── Calib0
│   │   ├── bias_correction
│   │   │   ├── SEbased
│   │   │   │   ├── AllGreyMatter.nii.gz
│   │   │   │   ├── [ .... ]
│   │   │   │   └── sebased_bias_dil.nii.gz
│   │   │   ├── gdc_sdc_calib0_bias.nii.gz
│   │   │   └── gdc_sdc_calib0_restore.nii.gz
│   │   ├── distortion_correction
│   │   │   ├── asl2orig_mgz_initial_bbr.dat
│   │   │   ├── [ .... ]
│   │   │   └── struct2asl.mat
│   │   ├── empirical_banding_correction
│   │   │   ├── eb_gdc_calib0.nii.gz
│   │   │   └── eb_gdc_sdc_calib0_restore.nii.gz
│   │   ├── aslfs_mask.nii.gz
│   │   ├── calib0.nii.gz
│   │   └── calib_timing.nii.gz
│   └── Calib1
│       ├── bias_correction
│       │   ├── SEbased
│       │   │   ├── AllGreyMatter.nii.gz
│       │   │   ├── [ .... ]
│       │   │   └── sebased_bias_dil.nii.gz
│       │   ├── gdc_sdc_calib1_bias.nii.gz
│       │   └── gdc_sdc_calib1_restore.nii.gz
│       ├── distortion_correction
│       │   ├── asl2orig_mgz_initial_bbr.dat
│       │   ├── [ .... ]
│       │   └── struct2asl.mat
│       ├── empirical_banding_correction
│       │   ├── eb_gdc_calib1.nii.gz
│       │   └── eb_gdc_sdc_calib1_restore.nii.gz
│       ├── aslfs_mask.nii.gz
│       └── calib1.nii.gz
├── perfusion_estimation # Basic perfusion estimates for BBR registration 
│   ├── logfile
│   ├── native_space
│   │   ├── aCBV.nii.gz
│   │   ├── arrival.nii.gz
│   │   ├── arrival_var.nii.gz
│   │   ├── mask.nii.gz
│   │   ├── perfusion.nii.gz
│   │   └── perfusion_var.nii.gz
│   └── oxford_asl_inputs
│       ├── beta_perf.nii.gz
│       └── brain_fov_mask.nii.gz
├── TIs # Intermediate ASL timeseries processing 
│   ├── bias_correction
│   │   ├── tis_restore.nii.gz
│   │   └── tis_dc_restore.nii.gz
│   ├── distortion_correction
│   │   └── FirstPass
│   │       ├── dc_tis.nii.gz
│   │       ├── temp_reg_dc_tis.nii.gz
│   │       └── tis_gdc.nii.gz
│   ├── empirical_banding_correction
│   │   ├── tis_dc_restore_eb.nii.gz
│   │   ├── tis_eb.nii.gz
│   │   ├── tis_eb_even.nii.gz
│   │   └── tis_eb_odd.nii.gz
│   ├── MoCo
│   │   ├── asln2m0.mat
│   │   │   ├── MAT_0000
│   │   │   ├── [ .... ]
│   │   │   └── MAT_0085
│   │   ├── asln2calibration_final.mat
│   │   │   ├── MAT_0000
│   │   │   ├── [ .... ]
│   │   │   └── MAT_0085
│   │   ├── final_registration_TIs.nii.gz
│   │   ├── final_registration_TIs.nii.gz.par
│   │   ├── fov_mask.nii.gz # Final motion-FoV mask 
│   │   ├── fov_mask_initial.nii.gz
│   │   ├── temp_reg_dc_tis_eb_odd.nii.gz
│   │   └── tis_dc_moco.nii.gz
│   ├── MotionSubtraction
│   │   ├── beta_baseline.nii.gz
│   │   ├── beta_perf.nii.gz
│   │   ├── combined_mask.nii.gz
│   │   └── difference_mask.nii.gz
│   ├── slicetime_correction
│   │   ├── st_scaling_factors.nii.gz
│   │   └── tis_stcorr.nii.gz
│   ├── slicetime_correction2
│   │   ├── combined_scaling_factors_asln.nii.gz
│   │   ├── st_scaling_factors.nii.gz
│   │   └── tis_dc_restore_eb_stcorr.nii.gz
│   ├── saturation_recovery
│   │   ├── nospatial
│   │   │   ├── finalMVN.nii.gz
│   │   │   ├── logfile
│   │   │   ├── mean_A.nii.gz
│   │   │   ├── mean_M0t.nii.gz
│   │   │   └── mean_T1t.nii.gz
│   │   └── spatial
│   │       ├── logfile
│   │       ├── mean_A.nii.gz
│   │       ├── mean_M0t.nii.gz
│   │       ├── mean_T1t.nii.gz
│   │       └── mean_T1t_filt.nii.gz
│   ├── saturation_recovery2
│   │   ├── nospatial
│   │   │   ├── finalMVN.nii.gz
│   │   │   ├── logfile
│   │   │   ├── mean_A.nii.gz
│   │   │   ├── mean_M0t.nii.gz
│   │   │   └── mean_T1t.nii.gz
│   │   └── spatial
│   │       ├── logfile
│   │       ├── mean_A.nii.gz
│   │       ├── mean_M0t.nii.gz
│   │       ├── mean_T1t.nii.gz
│   │       ├── mean_T1t_filt.nii.gz
│   │       └── mean_T1t_filt_asln.nii.gz
│   ├── brain_fov_mask.nii.gz # Combined brain and motion-FoV mask 
│   ├── brain_fov_mask_initial.nii.gz
│   ├── brain_mask.nii.gz
│   ├── combined_scaling_factors.nii.gz
│   ├── tis.nii.gz # Raw ASL timeseries 
│   ├── tis_dc_moco_restore.nii.gz
│   └── tis_dc_moco_restore_bandcorr.nii.gz # Fully corrected timeseries
├── gradient_unwarp
│   ├── fullWarp_abs.nii.gz
│   └── gdc_corr_vol1.nii.gz
├── topup
│   ├── Jacobian_01.nii.gz
│   ├── [ .... ]
│   ├── fmap_struct_reg
│   │   ├── asl2orig_mgz_initial_bbr.dat
│   │   ├── [ .... ]
│   │   └── asl2struct.mat
│   └── topup_params.txt
MNINonLinear # MNI space results 
├── ASL
│   ├── CIFTIPrepare
│   │   ├── arrival_Atlas.dscalar.nii
│   │   ├── [ .... ]
│   │   ├── perfusion_var_calib_MNI.nii.gz
│   │   └── pvcorr
│   │       ├── arrival_Atlas.dscalar.nii
│   │       ├── [ .... ]
│   │       └── perfusion_var_calib_MNI.nii.gz
│   ├── perfusion_estimation # Oxford ASL outputs (in MNI)
│   │   └── std_space
│   │       ├── arrival.nii.gz
│   │       ├── arrival_var.nii.gz
│   │       ├── perfusion_calib.nii.gz
│   │       ├── perfusion_var_calib.nii.gz
│   │       └── pvcorr
│   │           ├── arrival_gm_masked.nii.gz
│   │           ├── arrival_gm_var_masked.nii.gz
│   │           ├── arrival_wm_masked.nii.gz
│   │           ├── arrival_wm_var_masked.nii.gz
│   │           ├── perfusion_gm_calib_masked.nii.gz
│   │           ├── perfusion_gm_var_calib_masked.nii.gz
│   │           ├── perfusion_wm_calib_masked.nii.gz
│   │           └── perfusion_wm_var_calib_masked.nii.gz
│   ├── arrival.nii.gz # MNI NIFTI, ATT, non-PVEc
│   ├── arrival_Atlas.dscalar.nii # MNI CIFTI, ATT, non-PVEc
│   ├── arrival_cifti_mean_nonzero.txt # ROI calculation 
│   ├── perfusion_calib.nii.gz # MNI NIFTI, perfusion, non-PVEc
│   ├── perfusion_calib_Atlas.dscalar.nii # MNI CIFTI, perfusion, non-PVEc
│   ├── perfusion_calib_cifti_mean_nonzero.txt # ROI calculation 
│   ├── pvcorr_arrival_Atlas.dscalar.nii # MNI CIFTI, ATT, PVEc
│   ├── pvcorr_arrival_cifti_mean_nonzero.txt # ROI calculation 
│   ├── pvcorr_arrival_gm_masked.nii.gz # MNI NIFTI, ATT, PVEc GM
│   ├── pvcorr_arrival_wm_masked.nii.gz # MNI NIFTI, ATT, PVEc WM
│   ├── pvcorr_perfusion_calib_Atlas.dscalar.nii # MNI CIFTI, perfusion, PVEc
│   ├── pvcorr_perfusion_calib_cifti_mean_nonzero.txt # ROI calculation 
│   ├── pvcorr_perfusion_gm_calib_masked.nii.gz # MNI NIFTI, perfusion, PVEc GM
│   └── pvcorr_perfusion_wm_calib_masked.nii.gz # MNI NIFTI, perfusion, PVEc WM
T1w # T1w aligned space
└── ASL # T1w aligned at ASL resolution, used for final processing of all data. 
    ├── ASLQC # WB scene file and jupyter notebook for visual QC. 
    │   ├── HCD0378150_V1_MR_hcp_asl_qc.scene
    │   ├── HCD0378150_V1_MR_hcp_asl_qc_scene_1.png
    │   ├── HCD0378150_V1_MR_hcp_asl_qc_scene_2.png
    │   ├── HCD0378150_V1_MR_hcp_asl_qc_scene_3.png
    │   ├── HCD0378150_V1_MR_hcp_asl_qc_scene_4.png
    │   ├── HCD0378150_V1_MR_hcp_asl_qc_scene_5.png
    │   ├── HCD0378150_V1_MR_hcp_asl_qc_scene_6.png
    │   ├── HCD0378150_V1_MR_hcp_asl_qc_scene_7.png
    │   └── HCD0378150_V1_MR_hcp_asl_report.ipynb
    ├── CIFTIPrepare
    │   ├── arrival.L.badvert_ribbonroi.native.func.gii
    │   ├── [ .... ]
    │   ├── perfusion_var_calib_s2.atlasroi.R.32k_fs_LR.func.gii
    │   └── pvcorr
    │       ├── arrival.L.badvert_ribbonroi.native.func.gii
    │       ├── [ .... ]
    │       └── perfusion_var_calib_s2.atlasroi.R.32k_fs_LR.func.gii
    ├── Calib
    │   └── Calib0
    │       ├── distortion_correction
    │       │   ├── calib0_dc.nii.gz
    │       │   └── calib_mt_scaling_factors.nii.gz
    │       ├── SEbased
    │       │   ├── AllGreyMatter.nii.gz
    │       │   ├── [ .... ]
    │       │   └── sebased_bias_dilall.nii.gz
    │       ├── calib0_corrected.nii.gz
    │       ├── calib0_noncorr.nii.gz
    │       ├── calib_aslt1w_stcorr_factors.nii.gz
    │       └── calib_aslt1w_timing.nii.gz
    ├── HCD0378150_V1_MR_hcp_asl.log # HCP-ASL pipeline logfile 
    ├── perfusion_estimation
    │   ├── calib
    │   │   ├── M0.txt
    │   │   ├── logfile
    │   │   └── refmask.nii.gz
    │   ├── logfile
    │   ├── native_space
    │   │   ├── aCBV.nii.gz
    │   │   ├── [ .... ]
    │   │   ├── perfusion_wm_mean.txt
    │   │   ├── pvcorr
    │   │   │   ├── aCBV.nii.gz
    │   │   │   ├── [ .... ]
    │   │   │   └── perfusion_wm_wm_mean.txt
    │   │   └── wm_roi.nii.gz
    │   └── oxford_asl_inputs
    │       ├── beta_perf.nii.gz
    │       ├── brain_fov_mask.nii.gz
    │       ├── calib0_corrected.nii.gz
    │       ├── pve_GM.nii.gz
    │       ├── pve_WM.nii.gz
    │       ├── timing_img_aslt1w.nii.gz
    │       └── vent_csf_mask.nii.gz
    ├── PVEs
    │   ├── pve_GM.nii.gz
    │   ├── pve_WM.nii.gz
    │   └── vent_csf_mask.nii.gz
    ├── TIs
    │   ├── distortion_correction
    │   │   └── tis_dc_moco.nii.gz
    │   ├── MotionSubtraction
    │   │   ├── beta_baseline.nii.gz
    │   │   ├── beta_perf.nii.gz
    │   │   ├── combined_mask.nii.gz
    │   │   └── difference_mask.nii.gz
    │   ├── asl_corrected.nii.gz
    │   ├── asl_noncorr.nii.gz
    │   ├── combined_scaling_factors.nii.gz
    │   ├── reg
    │   │   ├── ASL_grid_T1w_brain_mask.nii.gz
    │   │   ├── [ .... ]
    │   │   ├── asl2struct.mat
    │   │   ├── brain_fov_mask.nii.gz
    │   │   ├── fmapmag_aslt1w.nii.gz
    │   │   ├── fov_mask.nii.gz
    │   │   └── mean_T1t_filt_aslt1w.nii.gz
    │   └── timing_img_aslt1w.nii.gz
    ├── aCBV_calib.nii.gz # T1w NIFTI, arterial cerebral blood volume, non-PVEc
    ├── arrival.nii.gz # T1w NIFTI, ATT, non-PVEc
    ├── arrival_gm_mean.txt # ROI calculation 
    ├── arrival_var.nii.gz
    ├── arrival_wm_mean.txt # ROI calculation 
    ├── asl_corrected.nii.gz # Fully corrected ASL timeseries 
    ├── asl_corr_subtracted.nii.gz # Fully corrected and subtracted ASL timeseries 
    ├── calib_corrected.nii.gz # Fully corrected calibration image 
    ├── perfusion_calib.nii.gz
    ├── perfusion_calib_gm_mean.txt
    ├── perfusion_calib_wm_mean.txt
    ├── perfusion_var_calib.nii.gz
    ├── pvcorr_aCBV_calib.nii.gz # T1w NIFTI, arterial cerebral blood volume, PVEc
    ├── pvcorr_arrival_gm_masked.nii.gz # T1w NIFTI, ATT, PVEc GM
    ├── pvcorr_arrival_gm_mean.txt # ROI calculation 
    ├── pvcorr_arrival_gm_var_masked.nii.gz
    ├── pvcorr_arrival_wm_masked.nii.gz # T1w NIFTI, ATT, PVEc WM
    ├── pvcorr_arrival_wm_mean.txt # ROI calculation 
    ├── pvcorr_arrival_wm_var_masked.nii.gz 
    ├── pvcorr_perfusion_calib_gm_mean.txt # ROI calculation 
    ├── pvcorr_perfusion_calib_wm_mean.txt # ROI calculation 
    ├── pvcorr_perfusion_gm_calib_masked.nii.gz # T1w NIFTI, perfusion, PVEc GM
    ├── pvcorr_perfusion_gm_var_calib_masked.nii.gz
    ├── pvcorr_perfusion_wm_calib_masked.nii.gz # T1w NIFTI, perfusion, PVEc WM
    ├── pvcorr_perfusion_wm_var_calib_masked.nii.gz
    ├── reg
    │   ├── ASL_grid_T1w_acpc_dc_restore.nii.gz
    │   └── wmmask.nii.gz
    └── roi_stats
        ├── fsl_identity.txt
        ├── roi_stats.csv
        ├── roi_stats_gm.csv
        └── roi_stats_wm.csv