Improved cardiovascular flow quantification with time-resolved volumetric phase-contrast MRI

Abstract

Background: Cardiovascular flow is commonly assessed with two-dimensional, phase-contrast MRI (2-D PC-MRI). However, scan prescription and acquisition over multiple planes is lengthy, often requires direct physician oversight and has inconsistent results. Time-resolved volumetric PC-MRI (4-D flow) may address these limitations.

Objective: We assess the degree of agreement and internal consistency between 2-D and 4-D flow quantification in our clinical population.

Materials and methods: Software enabling flow calculation from 4-D flow was developed in Java. With IRB approval and HIPAA compliance, 18 consecutive patients without shunts were identified who underwent both (1) conventional 2-D PC-MRI of the aorta and main pulmonary artery and (2) 4-D flow imaging. Aortic and pulmonary flow rates were assessed with both techniques.

Results: Both methods showed general agreement in flow rates (ρ: 0.87-0.90). Systemic and pulmonary arterial flow rates were well-correlated (ρ: 4-D 0.98-0.99, 2-D 0.93), but more closely matched with 4-D (P < 0.05, Brown-Forsythe). Pulmonary flow rates were lower than systemic rates for 2-D (P < 0.05, two-sample t-test). In a sub-analysis of patients without pulmonary or aortic regurgitation, 2-D showed improved correlation of flow rates while 4-D phase-contrast remained tightly correlated (ρ: 4-D 0.99-1.00, 2-D 0.99).

Conclusion: 4-D PC-MRI demonstrates greater consistency than conventional 2-D PC-MRI for flow quantification.

Figures

Figure 1

Cross-sectional images from the 4D flow acquisition, constructed at the plane of the aortic valve at peak-systole. The raw axial image is shown (a) with the location selected by the user to create a cross-section, labeled with an arrow. The software automatically creates a cross-section perpendicular to the direction of flow. The intersection of this plane with the axial image is marked by a blue solid line and the projection of the normal vector is displayed as a dashed line. The reformatted magnitude image (b), radial velocity (vr), rotational velocity (vθ) and through-plane velocity (vζ) are shown at peak-systole. The net flow is computed as the sum of the through-plane velocities in the segmented vessel lumen.

Figure 2

Representative flow curves (a) derived from 4D flow data from a patient post-repair of tetralogy of Fallot (same patient as figure 1). The red line with open circles shows systemic flow rates (Qs). The blue line with closed squares shows pulmonary flow rates (Qp) with pulmonary regugitation and diastolic flow reversal marked with an arrow. Left-anterior coronal oblique views are shown of aortic streamlines at peak-systole (b), pulmonary artery streamlines in early-systole (c) and pulmonary artery streamlines in early-diastole (d). Low-speed flow is colored in blue, intermediate-speed in yellow-green, and high-speed (150 cm/s) flow in red. Streamlines are shown overlaid on a translucent coronal plane. A recirculating current is seen in the main pulmonary artery during systole, marked by an arrow. Regurgitant flow is seen during diastole, also marked by an arrow.

Figure 2

Representative flow curves (a) derived from 4D flow data from a patient post-repair of tetralogy of Fallot (same patient as figure 1). The red line with open circles shows systemic flow rates (Qs). The blue line with closed squares shows pulmonary flow rates (Qp) with pulmonary regugitation and diastolic flow reversal marked with an arrow. Left-anterior coronal oblique views are shown of aortic streamlines at peak-systole (b), pulmonary artery streamlines in early-systole (c) and pulmonary artery streamlines in early-diastole (d). Low-speed flow is colored in blue, intermediate-speed in yellow-green, and high-speed (150 cm/s) flow in red. Streamlines are shown overlaid on a translucent coronal plane. A recirculating current is seen in the main pulmonary artery during systole, marked by an arrow. Regurgitant flow is seen during diastole, also marked by an arrow.

Figure 2

Representative flow curves (a) derived from 4D flow data from a patient post-repair of tetralogy of Fallot (same patient as figure 1). The red line with open circles shows systemic flow rates (Qs). The blue line with closed squares shows pulmonary flow rates (Qp) with pulmonary regugitation and diastolic flow reversal marked with an arrow. Left-anterior coronal oblique views are shown of aortic streamlines at peak-systole (b), pulmonary artery streamlines in early-systole (c) and pulmonary artery streamlines in early-diastole (d). Low-speed flow is colored in blue, intermediate-speed in yellow-green, and high-speed (150 cm/s) flow in red. Streamlines are shown overlaid on a translucent coronal plane. A recirculating current is seen in the main pulmonary artery during systole, marked by an arrow. Regurgitant flow is seen during diastole, also marked by an arrow.

Figure 2

Representative flow curves (a) derived from 4D flow data from a patient post-repair of tetralogy of Fallot (same patient as figure 1). The red line with open circles shows systemic flow rates (Qs). The blue line with closed squares shows pulmonary flow rates (Qp) with pulmonary regugitation and diastolic flow reversal marked with an arrow. Left-anterior coronal oblique views are shown of aortic streamlines at peak-systole (b), pulmonary artery streamlines in early-systole (c) and pulmonary artery streamlines in early-diastole (d). Low-speed flow is colored in blue, intermediate-speed in yellow-green, and high-speed (150 cm/s) flow in red. Streamlines are shown overlaid on a translucent coronal plane. A recirculating current is seen in the main pulmonary artery during systole, marked by an arrow. Regurgitant flow is seen during diastole, also marked by an arrow.

Figure 3

Bland-Altman comparison of 2D phase-contrast and 4D phase-contrast volumetric flow rate measurements using temporally-resolved segmentations. Relative differences in flow rates are shown. Measurements of systemic flow at the aortic valve are marked by circles and measurements of pulmonary flow at the pulmonary valve are marked by squares. The outer dashed-lines indicate the limits of agreement.

Figure 4

Bland-Altman comparison of pulmonary (Qp) and systemic (Qs) volumetric flow rate measurements by (a) 2D phase-contrast and (b) 4D phase-contrast MRI using temporally-resolved segmentations. The outer dashed-lines indicate the limits of agreement. Note marked improvement with 4D phase-contrast, showing tighter limits and a mean difference between Qp and Qs closer to zero.

Figure 4

Bland-Altman comparison of pulmonary (Qp) and systemic (Qs) volumetric flow rate measurements by (a) 2D phase-contrast and (b) 4D phase-contrast MRI using temporally-resolved segmentations. The outer dashed-lines indicate the limits of agreement. Note marked improvement with 4D phase-contrast, showing tighter limits and a mean difference between Qp and Qs closer to zero.