Robust estimation of quantitative perfusion from multi-phase pseudo-continuous arterial spin labeling

Area of Science:

• Medical Imaging
• Neuroscience
• Biophysics

Background:

• Multi-phase pulsed arterial spin labeling (PCASL) is used for accurate perfusion quantification.
• A known bias in PCASL estimation is linked to data noise and imperfect shim.
• Addressing this bias is crucial for reliable perfusion measurements.

Purpose of the Study:

• To characterize and correct the bias in multi-phase PCASL perfusion quantification.
• To develop an algorithm that overcomes noise-induced bias in perfusion estimation.
• To validate the algorithm in both preclinical (rat) and clinical (human) datasets.

Main Methods:

• A novel algorithm utilizes biased voxel-wise estimates to cluster phase shifts and derive high-SNR regional phase offsets.
• Simulations predicted bias at typical signal-to-noise ratios (SNR).
• The method was applied to multi-phase PCASL data from rats (n=21) at 9.4 T and humans (n=20) at 3 T, including arterial blood flow velocity estimation.

Main Results:

• Simulations indicated a 6-8% perfusion estimation bias at typical SNR with 8-phase data.
• The bias was eliminated with the high-precision phase error estimate.
• Bias-corrected perfusion in rats was 107 ± 14 mL/100g/min (8.0% bias reduction), and human data showed a 15% decrease in perfusion estimates.

Conclusions:

• A retrospective algorithmic approach effectively utilizes regional information to enhance SNR.
• This method overcomes bias in multi-phase PCASL perfusion quantification.
• The developed technique offers improved accuracy for perfusion imaging in both research and clinical settings.