VPAL: A novel method to reduce reconstruction time for 5D free-running imaging

TL;DR

The paper addresses the lengthy reconstruction time of ferumoxytol-enhanced 5D free-running whole-heart CMR by introducing VPAL, a variable projection augmented Lagrangian method, and compares it to the standard ADMM approach. VPAL uses a projected descent update with a nonlinear Fletcher-Reeves conjugate gradient to accelerate convergence while maintaining image quality and functional metrics such as LVEF. Across numerical simulations and 15 pediatric in-vivo datasets, VPAL achieved comparable accuracy to ADMM but with substantially reduced reconstruction times (roughly 70% faster in vivo and ~27% faster in simulations), and preserved sharpness, SSIM, and radiologist-diagnostic ratings. This work suggests VPAL can enable clinically feasible 5D free-running reconstruction workflows, potentially reducing scan-time uncertainty and improving workflow for complex cardiac imaging.

Abstract

Purpose: Ferumoxytal-enhanced 5D free-running whole heart CMR provides image quality comparable to CTA, but requires hours-long reconstruction time, preventing clinical usage. This study developed a variable projection augmented Lagrangian (VPAL) method for 5D motion-resolved image reconstruction and compared it with alternating direction method of multipliers (ADMM) in five numerical simulations and 15 in-vivo pediatric data set. Approach: Relative error of the reconstructed images against the ground-truth images was assessed in numerical simulations. In-vivo analysis compared reconstruction time, mid-short axis (SA) blood-myocardium sharpness, left ventricular ejection fraction (LVEF), and a radiologist's image quality ratings between VPAL and ADMM. A paired t-test (p<0.05) was used to determine statistical significance, while linear regression and Bland-Altman analysis for agreement assessments. Results: VPAL and ADMM had similar relative errors compared to the ground truth, p = 0.07. In in-vivo datasets, VPAL reduced the reconstruction time from 16.3 +/- 3.6 hours (ADMM) to 4.7 +/- 1.1 hours (VPAL), p=1e-10. Blood-myocardium border sharpness in VPAL closely correlates to ADMM , R^2 = 0.97. The LVEFs values measured by VPAL and ADMM reconstructions are largely similar, 56 +/- 6 % in ADMM and 56 +/- 6 % in VPAL, p=0.55. Both VPAL and ADMM reconstructions have good to excellent diagnostic ratings (VPAL vs. ADMM: 3.9 +/- 0.3 vs. 3.8 +/- 0.4 in 2-chamber; 3.9 +/- 0.4 vs. 3.9 +/- in 4-chamber; 3.7 +/- 0.5 vs. 3.7 +/- 0.5 in mid-SA reformatted views. Conclusion: VPAL enables faster reconstruction than ADMM while maintaining equivalent image quality for functional assessments, supporting its potential for clinical use.

Figures (8)

• Figure 1: Example numerical simulation reconstructions The 4-chamber, mid-short-axis, and 2-chamber view in cardiac resting phase demonstrated in the ground truth 5D image, and the output of the reconstructions using ADMM and VPAL algorithms.
• Figure 2: Boxplot of convergence time using ADMM and VPAL reconstruction algorithms.No statistical significance is found on the convergence time using ADMM versus VPAL
• Figure 3: Linear fitting of the relative error between ADMM and VPAL reconstructions with the ground-truth
• Figure 4: Exampe in-vivo reconstructions using ADMM and VPAL on three subjects The mid-short-axis, 2-chamber, and 4-chamber view in cardiac resting phase demonstrated in the output of the reconstructions using ADMM and VPAL algorithms.
• Figure 5: Boxplot of reconstruction time using ADMM and VPAL algorithms. On 15 in-vivo datasets, VPAL reconstruction is much faster than ADMM reconstruction.