Purpose

The purpose of this study is to verify the accuracy of the measurements of CS-based flow velocity in 3D cine phase-contrast MRI, using a transparent flow phantom and a steady-flow pump.

Methods

Two types of phantoms were used to verify the accuracy of the measurements of the flow velocity. The phantom (A) was an acrylic tube with an inner diameter of 7 mm, installed along the diagonal of an acrylic container measuring 10 cm on each side (fig. 1). The container was fixed with 2% agarose containing a 0.1 mmol/L Gd-based contrast agent. Each of the phase images obtained in the anterior-posterior (A-P), the foot-head (F-H), and the right-left (R-L) directions captured 1/√3 of the actual flow velocity. In theory, the combined flow velocity of the phase images in the three directions should be equal to the actual flow velocity. A glycerin solution diluted with distilled water was used as the simulated blood in the acrylic tube. The fluid was circulated using a steady-flow pump, and the average flow velocity was adjusted to 30 cm/s (fig. 2). The phantom (B) was an acrylic tube with an inner diameter of 8 mm, installed in an acrylic container measuring 10 cm on each side (fig. 3). To meet the conditions for a laminar flow, the Reynolds number was kept below 2,100, and a duct of sufficient length was secured in the passage of the flow. The pump was adjusted to achieve average flow velocities of 20, 40, and 60 cm/s. The flow rates were measured using a flowmeter, and the resulting values were divided by the cross-sectional area of the acrylic tubes to obtain the baseline (actual) average flow velocities.
MR images were obtained using a 3T MR scanner (Ingenia Elimino X, Philips healthcare, Best, the Netherlands). The images were captured by the 3D cine phase-contrast MR imaging (also referred to as the “4D-flow” method) using simulated electrocardiography. Measurements were taken using three different techniques of image time reduction: CS, sensitivity encoding parallel imaging (SENSE), and non-acceleration. The average and maximum velocities obtained using each technique were compared to the actual velocities.

Results

Fig. 4 shows the average velocities of the simulated blood in the three directions, and the corresponding maximum velocities are shown in Fig. 5. The application of each of the three techniques of image time reduction resulted in slightly higher flow velocities in the A-P direction. Furthermore, slightly higher flow velocities were observed when the SENSE technique was applied compared with the other two techniques, though there were no significant differences. Fig. 6 illustrates the average velocities when the flow rate was varied, and the corresponding maximum velocities are depicted in Fig. 7. The horizontal axis represents the actual measurements, while the vertical axis represents the flow velocities measured in the MRI. In case all three techniques, the average flow velocity was highly correlated with the measured value, though the maximum velocity was overestimated in proportion to the increase in the flow rate.

Consideration

It has been found in previous studies that high spatial resolution is required to accurately measure the average and maximum flow velocities in blood vessels. This finding was taken into account in our study, and the necessary spatial resolution was set. Although, the maximum velocity declines due to partial volume effects, yet it tends to be higher than the baseline value. This might be partially due to the Reynolds number of the 40% glycerin solution used as the simulated blood.

Conclusions

In this study, we were able to verify the accuracy of the measurements of CS-based flow velocities in 3D cine phase-contrast MRI, using two phantoms and a steady-flow pump. The simulated blood flowing across the diagonal yielded nearly equal flow velocities in all three directions, and this suggested that it should be possible to calculate the flow velocity vectors, with accuracy, at any point. Furthermore, no significant difference was observed between the CS, the SENSE, and the non-acceleration techniques, leading to conclude that all three enable the flow velocity to be measured with equal accuracy.

Acknowledgements

We would like to express our gratitude to Department of Radiology, Mie University Hospital for giving us permission to use their MR imaging scanner. This work was supported by JSPS KAKENHI Grant Number 19K12863.