Investigation of 4D flow in high heart rates using an aorta flow phantom

Background: After successful repair of coarctation of the aorta, some patients suffer from hypertension and reduced exercise capacity. However, the pathophysiology is not fully understood (1). Four-dimensional flow imaging (4D flow) during exercise CMR could be a useful technique for investigating blood flow in these patients with the aim of explaining the pathophysiology. Compressed Sensing (CS) 4D flow has been shown to reduce scan time in aortic CMR (2) and has been validated in healthy volunteers (2), adult (3) and pediatric patients (4) at rest. However, it has not been evaluated in high heart rates as seen during exercise CMR where the limited temporal resolution in typical 4D flow protocols may lead to significant quantification errors.

Therefore, we aimed to develop and test a flow phantom platform for investigating temporal resolution in 4D flow sequences at high heart rates such as seen in exercise CMR.

Methods: An adolescent-sized (approximately 10 years old) aorta was 3D printed (Figure 1). The phantom included an imaging plane where all vessel diameters were equal and larger than the ones in the anatomical positions, which was used to collect reference flow values (reference plane).

The phantom was connected to a custom pulsatile flow pump (developed by the authors) which was driven by a motion-transfer cable (DZ95, Flexball, Orbassano, Italy), connected to an servo-driven actuator (CM2-C-60A40C-K, Myostat Motion Control, Newmarket, Canada) and simulated conditions of a) rest (75 bpm) and b) light exercise (120 bpm). The cardiac output was 6 l/min in both cases to isolate the effect of heart rate in flow quantification.

Scans were performed using a 1.5T clinical scanner (MAGNETOM Aera, Siemens Healthineers, Forchheim, Germany). 4D flow using CS acceleration was performed over the whole phantom using a research sequence. Additionally, 2D flow was performed at the reference plane. Table 1 shows pulse sequence parameters.

Results: Figure 2A shows example images as acquired with CS 4D flow at 120 bpm. Flow curves from the AAo with 2D flow at the reference plane and with 4D flow at the anatomical position can be seen in Figure 2B. Correlation (75 bpm: R=0.99, 120bpm: R=0.99) and modified Bland-Altman plots of net flow measured in the anatomical position of each vessel are shown in Figure 3. Compared to the 2D acquisition, CS 4D flow bias± SD at 75 bpm was -0.17±0.20 l/min and -0.76±0.28 l/min at 120 bpm.

Conclusion: We developed an in-house setup for flow phantom experiments with high heart rates such as seen in exercise CMR. The setup can be used to investigate temporal resolution of 4D flow sequences for protocol optimization for exercise flow imaging.

Pulse sequence parameters

Parameter	2D	4D
Acceleration	none	Compressed sensing (R=7.7)
Segmentation factor	1	2
TR (ms) /TE (ms) /Flip angle (degrees)	4.9/ 2.7 / 20	4.8/2.3/8
Matrix size (read x phase x slice mm3)	400 x 323 x 5	96x96x56
Spatial resolution acquired (read x phase x slice mm3)	1.92 x 1.92 x 5	2.5 x 2.5 x 2.5
Spatial resolution reconstructed (read x phase x slice mm3)	1.92 x 1.92 x 5	2.5 x 2.5 x 2.5
Temporal resolution acquired (ms)	9.8	38.4
Temporal resolution reconstructed (ms)	14.2	21.2
VENC (cm/s)	300	300
Scan time (sec)	126	140

Aorta flow phantom description
Example images and flow curves as acquired with CS 4D flow at 75bpm and 120 bpm
Correlation (left) and modified Bland-Altman (right) plots to compare flow quantification with CS 4D and 2D flow