Dual-Venc acquisition for 4D flow MRI in aortic stenosis with spiral readouts

Abstract

Background: Single V_enc 4D flow MRI with Cartesian readout is hampered by poor velocity resolution and noise when imaging during diastole. Dual V_enc acquisitions typically require the acquisition of two distinct datasets, which leads to longer scan times.

Purpose/hypothesis: To design and develop a 4D Spiral Dual V_enc sequence. The sequence allows for separate systolic and diastolic V_enc s as part of a single acquisition with a prescribed switch time. The implemented sequence was hypothesized to be comparable to Cartesian 4D flow, but with increased velocity resolution in the diastolic phase and with better scan efficiency and reduced noise.

Study type: Prospective.

Population: The studied populations were two phantoms-a straight pipe with a stenotic narrowing and a phantom of the aortic arch which included a calcific polymeric valve-under both steady and pulsatile flows, six healthy volunteers, and eight patients with severe aortic stenosis (AS).

Field strength/sequence: 1.5T, Dual V_enc 4D flow with spiral readouts.

Assessment: Data from the proposed sequence were compared with data from 4D Cartesian Dual V_enc and Single V_enc acquisitions. Noise was assessed from the acquired velocity data with the pump turned off and by varying V_enc . Steady acquisitions were compared to the proximal slice of the lowest Single V_enc acquisition.

Statistical tests: Steady flows were compared using relative-root-mean-squared-error (RRMSE). For in vivo flows and pulsatile in vitro flows, net flow for corresponding timepoints were compared with the Pearson correlation test (P < 0.01).

Results: For steady flows, RRMSEs for Single V_enc s ranged from 17.6% to 19.4%, and 9.6% to 16.5% for Dual V_enc s. The net flow correlation coefficient for the aortic arch phantom was 0.975, and 0.995 for the stenotic phantom. Normal volunteer and patient comparisons yielded a correlation of 0.970 and 0.952, respectively. in vitro and in vivo pulsatile flow waveforms closely matched.

Data conclusion: The Dual V_enc offers improved noise properties and velocity resolution, while the spiral trajectory offers a scan efficient acquisition with short echo time yielding reduced flow artifacts.

Level of evidence: 2 Technical Efficacy Stage: 1 J. Magn. Reson. Imaging 2020;52:117-128.

Keywords: 4D flow MRI; Dual Venc; non-Cartesian trajectory; phase contrast MRI; spiral acquisition; stenotic flow.

FIGURE 1:

The Dual V_enc pulse sequence. The bipolar gradients change the maximum based on the set V_enc. The V_enc of the sequence on the left is higher than the one on the right. The switch to the low V_enc happens at T_DV, which is a user defined time in the R-R interval.

FIGURE 2:

(a) The flow circuit. The open-air reservoir holds the blood mimicking fluid, which circulates through the system. The pump pulls the fluid from the reservoir, through a flexible plastic tubing. The pump then pushes the fluid in the flow circuit as prescribed by the programmed flow waveform. The fluid then exits the phantom and returns to the reservoir. (b) A picture of the stenotic phantom on the MRI table; the acrylic phantom has a 1” inner diameter, which narrows to 0.35” at the throat for an 87% area occlusion. (c) The aortic arch phantom precision machined with 1” diameter from clear acrylic. The setup permits changing polymeric valves in a separate block close to the inlet. (d) Calcific polymeric valves, at 50% (left) and at 0% (right) calcification. The valves, which are 1” in diameter, are made of polyurethane with various levels of calcium phosphate.

FIGURE 3:

(a) Velocity magnitude map of a sagittal cut of the stenotic phantom distal to the throat or region of interest during systole. Q_max = 150 ml/s (range: 100.0 cm/s to 0.0 cm/s. (b) Velocity magnitude map in diastole for the same study (range: 100.0 cm/s to 0.0 cm/s). (c) Velocity magnitude map of the aortic arch phantom with a 0% calcification synthetic valve at peak systole (Q_max = 150 ml/s). The valve is located on the lower left portion of the picture; the high-speed jet may be seen immediately distal to the valve (range: 100.0 cm/s to 0.0 cm/s). (d) The same sagittal slice as in (c), but during diastole (range: 100.0 cm/s to 0.0 cm/s). (e) peak systolic axial slice of velocity magnitudes corresponding to (c) and immediately distal to the valve. The valve opening clearly visible with its trileaflets (range: 175.0 cm/s to 0.0 cm/s). (f) Corresponding diastolic axial slice of velocity magnitudes (range: 50.0 cm/s to 0.0 cm/s).

FIGURE 4:

No-flow Single V_enc results. Four sequences are compared. These are: Spiral High V_enc (200 cm/s), Spiral Low V_enc (40 cm/s), Cartesian High V_enc (200 cm/s), and Cartesian Low V_enc (40 cm/s). (a) flow error by slice. (b) Mean and standard deviation of velocity error as a function of Slice 5 is approximately at the isocenter and is at the stenosis throat.

FIGURE 5:

(a) Flow waveform of the aortic arch phantom (Fig. 2c) with a synthetic valve, with 0% occlusion. The slice chosen was 5 mm distal to the valve. (b) Flow waveform in the stenotic phantom (Fig. 2b) calculated from a slice location 4 mm distal to the stenosis throat. Note the Spiral Single V_enc waveform is artifactual because of velocity aliasing. In this figure the Single V_enc acquisitions are plotted with dotted line, while the Dual V_enc results are plotted with a solid line. The Cartesian has a data marker of an x, while Spiral has a circular marker. (c) Comparison of Dual V_enc Spiral and Dual V_enc Cartesian flow data points for the slice immediately distal to the 50% polymeric valve in the aortic arch phantom. The Pearson correlation coefficient for the scatterplot is 0.975. (d) Comparison of the same Dual V_enc Spiral, and High V_enc Cartesian in the stenotic phantom. Pearson correlation coefficient for this scatterplot is 0.995. See text for additional description.

FIGURE 6:

(a) Flow waveform in a normal subject distal to the aortic valve from Spiral and Cartesian Single V_enc and Spiral Dual V_enc acquisitions. (b) Waveforms measured with the same sequences in a different normal subject. (c) Scatterplot of flow data for all healthy volunteers (n = 6, Pearson correlation coefficient of 0.970). (d) Scatterplot of flow data for all patients with severe aortic stenosis (n = 7, Pearson correlation coefficient of 0.952). For both (c,d), flow rates for a specific slice, timepoint, and volunteer are single points.

FIGURE 7:

(a) Flow waveform in a patient with severe AS as measured with a Dual V_enc, a high V_enc and a low V_enc spiral acquisition for the entire R-R interval. The T_DV was set to 400 msec. Note that the low V_enc acquisition leads to velocity aliasing in systole and an artifactual flow waveform. (b) The scatterplot of the Dual V_enc vs. High V_enc acquisition in the same patient. All slice positions and timepoints were included. The Pearson correlation coefficient was 0.930, (c) (range: 150 cm/s to 0.0 cm/s) and (d) (range: 25.0 cm/s to 0.0 cm/s) show magnitude of velocities in an axial slice at the aortic valve level in the same patient at peak systole and in mid-diastole.