ASL Cerebral Blood Flow Workflow

An end-to-end workflow estimating cerebral blood flow (CBF) metrics from arteral spin labeling (ASL) MRI data. ASL is a novel brain perfusion technique which enables the quantification of CBF non-invasively, without ionizing radiation and without the use of exogenous contrast agents, rendering it useful in investgations of cerebrovascular diseases and neurodegenerative diseases alike.

Based on the BASIL toolset available within FSL, the pipeline computes both CBF and CBF-calibrated maps, the latter with units of ml/100g/min.

The full workflow includes:

  • Image preprocessing, including motion correction and partial volume estimation.
  • Registration to FreeSurfer space.
  • Automatic extraction of CBF statistics for various ROIs, including:
    • WM/GM/CSF
    • Cerebellar gray and white matter
    • Caudate
    • Putamen
    • Pallidum

NOTE: This workflow is currently optimized to work only with certain GE and Siemens pcASL sequences.

Report output examples:

Required inputs:

  • ASL image
    • Arterial Spin Labelling 4D series
    • Must have modality 'ASL'
    • All label and control acquisitons must be paired.
  • FreeSurfer recon-all segmentation output
    • Segmentation performed previously using a T1 image of the same individual and imaging session.

Minimum input requirements:

  • For optimal result reliability, ASL resolutions should be low.
  • ASL recommended resolution: from [2.5 x 2.5 x 4] to [3.5 x 3.5 x 6]

Optional inputs:

  • Proton density-weighted or perfusion-weighted image
    • Must have modality 'PD' or modality 'SCALAR' and tag 'asl'

Settings:

  • ASL acquisition type
    • Choose from:
      • cASL
      • pcASL
      • pASL
  • MRI acquisition type
    • Choose from:
      • 2D
      • 3D
  • ASL data format
    • Choose from:
      • Label-Control
      • Control-Label
      • Difference
  • Scanner manufacturer
    • Choose from:
      • GE
      • Siemens
      • Philips
      • Other
  • Repetition Time PD [s]
  • Time Between Slices [s]
  • Post Labeling Delay [s]
  • Bolus Duration [s]

Outputs:

  • Report:
    • report.pdf: report file with results summary.
  • Images:
    • ASL-in-fs.nii.gz: original ASL image in FreeSurfer space
    • ASL.nii.gz: original ASL image
    • PD.nii.gz: original PD image
    • CBF/*: scalar output maps
    • aux/*: anatomical (segmented) images
  • Summary:
    • ASLCBF_{subj}_{ssid}_ROI_measurements.csv: CSV of CBF summary statistics per ROI
    • ROI_measurements_info.csv : extended summary statistics per ROI (+_in-asl derived from lower resolution ASL space)
    • ROI_values-in-asl.csv: per ROI voxel-wise CBF values (+_in-asl derived from lower resolution ASL space)

Typical execution time: ~1 hour

References:

  • FSL's BASIL:
    Chappell MA, Groves AR, Whitcher B, Woolrich MW. Variational Bayesian inference for a non-linear forward model. IEEE Transactions on Signal Processing 57(1):223-236, 2009.