Reference-Superimposed Reconstruction (RS-Recon) for Arterial Spin Labeling

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)

Background:

• Conventional arterial spin labeling (ASL) reconstruction faces challenges with sign-flipping errors and noise bias, particularly under strong background suppression (BS).
• These limitations hinder signal-to-noise ratio (SNR) optimization in ASL imaging.

Purpose of the Study:

• To introduce and evaluate reference-superimposed reconstruction (RS-recon) as a novel method for ASL image reconstruction.
• To enable ideal background suppression and robust signal recovery, overcoming limitations of conventional methods.

Main Methods:

• RS-recon involves superimposing high-SNR M₀ k-space data onto ASL k-space before reconstruction, followed by reference subtraction.
• The method was tested on four volunteers using pulsed ASL, pseudo-continuous ASL, and dual-module velocity-selective ASL at 3T across varying BS levels.
• RS-recon integrated pipelines were compared against standard magnitude, complex subtraction, and complex reconstruction pipelines using metrics like ASL signal, temporal SNR (tSNR), and artifact levels.

Main Results:

• RS-recon effectively eliminated subtraction artifacts caused by signal sign-flipping and recovered negative ASL signals.
• Significant improvements in ASL signal and tSNR were observed in gray and white matter across all ASL methods and BS conditions.
• The method demonstrated robustness against misalignments, preserved signal fidelity, and integrated well with generalized auto-calibrating partially parallel acquisitions (GRAPPA).

Conclusions:

• RS-recon provides a simple and robust strategy for ASL image reconstruction, resolving subtraction errors and enabling optimal background suppression.
• The technique enhances signal fidelity, SNR, and phase accuracy.
• RS-recon integrates seamlessly with existing reconstruction pipelines and parallel imaging techniques.