Introduction

Time-resolved 3-directional 2D PC-MRI and its extension to 3D, 4D flow MRI, enable non-invasive visualization of blood flow patterns and quantification of hemodynamic parameters, providing valuable diagnostic information for various cardiovascular diseases (1). PC-MRI and especially 4D flow MRI have been under continuous development and optimization for instance to improve spatial resolution, dynamic velocity range, and acquisition time. During development process, MRI phantoms are indispensable tools for examining new algorithms and evaluating accuracy and reproducibility of MR sequences for velocity measurement (2,3). It is also useful to regularly evaluate the quality of MR images using a reference phantom to ensure consistency across MR scanners for clinical studies. An ideal phantom should establish well-defined velocity fields, so the measured velocities can be compared with the ground truth, and it should reproduce the same velocities under the same settings. In this study, an air-driven rotation phantom was designed to meet these criteria, and its performance was examined.

Methods

The phantom consists of a cylinder with a diameter of 128 mm surrounded by a ring with inner and outer diameters of 200 and 250 mm (Figure 1). The cylindrical component rotates with a known rotational speed, while the ring remains static providing a stationary reference for phase offset correction. Both components of the phantom were filled with agarose gel with added sodium azide, NaN3, to protect against spoilage and gadolinium at 0.2% volumetric ratio to improve MR contrast. The top of the cylinder was designed as a centrifugal impeller with 12 blades. A nozzle creates a jet of pressurized air pushing against the impeller blades to rotate the cylinder. The rotational speed in revolutions-per-minute (RPM) was measured and recorded in real-time with an optical counter consisting of a photosensor that generates voltage pulses when light from the sender to receiver of the sensor is disrupted by an obstacle mounted underneath the cylinder. Using a valve, the air pressure from a tank was controlled to accomplish desired rotational speeds. The phantom was placed in the isocenter of a 1.5T MR scanner (Aera, Siemens, Germany) where it was rotating in the coronal plane. Once the phantom had constant rotational speed, 3-directional 2D PC-MRI scans with three different velocity sensitivity (VENC) for three rotational speeds (100, 150, and 200 RPM) were acquired (Siemens PC-MRI sequence): TR/TE=5.9-6.2/2.5-2.7 ms, flip angle=15°, voxel=2 mm isotropic. VENCs of 90, 120, and 150 cm/s were used based on expected maximum velocities of 67, 100, and 134 cm/s, respectively. For phase offset error (eddy current) correction, the static ring was identified in magnitude images, a second-order polynomial was fitted to background phase in static regions and was subtracted from the acquired data in each velocity direction separately. Knowing the linear relationship of the velocity magnitude with the distance from the center of the cylinder, the velocity components were calculated analytically and compared with the velocity components from PC-MRI. For test-retest experiment, the phantom rotating at 150 RPM was imaged with the same sequence parameters on two different days.

Results

Phase images from 2D PC-MRI acquisition of the phantom rotating at 100 RPM were used to calculate 2D maps of velocity components and velocity magnitude which are compared with the analytical velocity maps in Figure 2. Velocity values were sampled across the diameter of the cylinder radially every 5° as shown in Figure 2e., resulting in 36 values for the velocity magnitude at each point across the phantom. Mean and 95% confidence interval of velocity magnitude at different points along the diameter of the phantom are presented in Figure 3 and are compared with the analytical solution at the three different rotational speeds. The intraclass correlation coefficient (ICC) of velocity magnitude were 0.999, 0.999, and 0.997 for 100, 150, and 200 RPM, respectively. The velocity magnitude at different locations along the diameter of the phantom rotating at 150 RPM on two different days is presented in Figure 4. The ICC for the test-retest comparison was 0.999.

Discussion

A pressurized air-driven rotation phantom with optical real-time feedback of rotational speed was developed to be used to evaluate the accuracy in measuring tissue motion or blood flow velocity with PC-MRI sequences. The phantom provides a well-defined and continuous linear velocity distribution which can be used as ground truth velocity field allowing quantitative error analysis. In addition, the velocity field created by the phantom was reproducible, which makes the phantom a reliable reference for sequence validation and quality control.

Acknowledgements

Financial support by Siemens Healthcare, AHA 16SDG30420005, NIH R01HL115828