Quantification of thoracic blood flow using volumetric magnetic resonance imaging with radial velocity encoding: In vivo validation

Objectives: The objective of this study was to validate radially undersampled 5-point velocity-encoded time-resolved flow-sensitive magnetic resonance imaging (MRI) ("PC-VIPR", phase contrast vastly undersampled imaging with isotropic resolution projection reconstruction magnetic resonance) for the quantification of ascending aortic (AAO) and main pulmonary artery (MPA) flow in vivo. Materials and Methods: Data from 18 healthy volunteers (41.6 ± 16.2 years [range, 22-73 years]; body mass index, 26.0 ± 3.5 [19.1-31.4]) scanned at 3 T with a 32-channel coil were included. The left and right ventricular stroke volumes calculated from contiguous short-axis CINE-balanced steady state free precession (CINE-bSSFP) slices were used as the primary reference for cardiac output. Flow measured from 2-dimensional phase contrast MRI (2D-PC-MRI) in the AAO and the MPA served as the secondary reference. Time-resolved 4-dimensional flow-sensitive MRI (4D flow MRI) using PC-VIPR was performed with a velocity sensitivity of V_enc = 150 cm/s reconstructed to 20 time frames at 1.4-mm isotropic spatial resolution. In 11 of 20 volunteers, phantom-corrected 4D flow MRI data were also assessed. Differences between methods of calculating the left ventricular and right ventricular cardiac output were assessed with the Bland-Altman analysis (BA, mean difference ±2SD). The Q_P/Q_S-ratio was calculated for each method. Results: Initially, PC-VIPR compared unfavorably with CINE-bSSFP (left ventricular stroke volume: 96.5 ± 14.4 mL; right ventricular stroke volume: 93.6 ± 14.0 mL vs 81.2 ± 24.3 mL [AAO] and 85.6 ± 25.4 mL [MPA]; P = 0.027 and 0.25) with BA differences of -14.6 ± 44.0 mL (AAO) and -9.0 ± 45.9 mL (MPA). Whereas phantom correction had minor effects on 2D-PC-MRI results and comparison with CINE-bSSFP, it improved PC-VIPR results: BA differences between CINE-bSSFP and PC-VIPR after correction were -1.4 ± 15.3 mL (AAO) and -4.1 ± 16.1 mL (MPA); BA comparison with 2D-PC-MRI improved to -12.0 ± 48.1 mL (AAO) and -2.2 ± 19.5 mL (MPA). Q_P/Q_S-ratio results for all techniques were within physiologic limits. Conclusions: Accurate quantification of AAO and MPA flows with radially undersampled 4D flow MRI applying 5-point velocity encoding is achievable when phantom correction is used.