1354400 - Spatial resolution impacts 4D Flow CMR parameters

4D Flow Cardiovascular Magnetic Resonance (CMR) allows comprehensive haemodynamic assessment. High spatial resolution improves image quality and accuracy but requires longer image acquisition time. In congenital cardiology, with a large paediatric population, prolonged scan duration may impact the feasibility of high resolution 4D Flow CMR in clinical practice.

Aim

To explore how spatial resolution impacts 4D Flow CMR parameters.

Methods:

Twenty volunteers were prospectively recruited (11 males, median age 24 years [range 17-33 years], median weight 73kgs (range 54-127kgs). The first 10 consecutive volunteers were imaged on a 3T clinical scanner (MAGNETOM Prisma 3T, Siemens Healthcare, Erlangen, Germany) using standard 4D flow CMR sequence at resolutions of 2mm³ and 3mm³, and the second 10 consecutive volunteers were imaged using compressed sensing 4D Flow CMR (CS 4D Flow CMR) at resolutions 1.5mm³ (aorta only), 2mm³, 3mm³ and 4mm³ (Figure 1). All volunteers also underwent 2D phase contrast (PC) assessment of ascending aortic (AAo), main pulmonary artery (MPA) and branch pulmonary arteries (BPAs).

Standard analysis of flow volume was performed using commercially available software (PIE medical imaging software, CASS, The Netherlands). Advanced parameters were derived using research-based software solutions.

Spearman’s Rho was performed for correlation analysis due to small sample sizes. Bland-Altman analysis was used to assess agreement.

Results:

AAo, MPA and combined BPAs forward flow showed a positive correlation between standard, validated 2D PC and standard 4D Flow CMR acquired at 2mm³ and 3mm³ in both subgroups and additionally at 1.5mm³ in subgroup 2. However, CS 4D Flow CMR at 4mm³ resolution did not significantly correlate with standard 2D PC (Table 1).

For the AAo, MPA and BPAs CS 4D Flow CMR at 2mm³ showed the least biases with the best limits of agreement; especially for 4mm³ limits of agreement were unacceptably large (Table 2).  

For advanced intra-cardiac measures, left ventricular (LV) peak E-wave and peak systolic kinetic energy (KE) showed strong correlation between 2 and 3mm³ for both 4D flow CMR acquisitions. Data acquired at 4mm³ did not correlate for any parameters (Table 1). Limits of agreement were smaller for the standard 4D Flow CMR sequence (Table 2).

For advanced aortic measures (KE, vorticity, helicity, energy loss, wall shear stress, viscous dissipation) all 4 spatial resolutions did not correlate well with each other.

Conclusion:

Standard and CS 4D Flow CMR resolutions up to 3mm³ are comparable to standard and validated 2D PC for clinical use. For intra-cardiac KE measures 2 and 3mm³ resolutions correlate. However, 4mm³ correlates poorly across all ventricular parameters and should not be used. Advanced aortic measures are most heavily impacted by different spatial resolutions and consistent use of similar spatial resolution is important to achieve comparable results across cohorts.