Objective: Differences of opinion in the literature about the agreement between magnetic resonance imaging (MRI) and ultrasound Doppler (USD) for high blood velocity measurement were the basis for this investigation. It compares in vitro velocity measurements by MRI with intraluminal USD, using a specially designed flow phantom.
Design: The flow phantom consisted of a pulsatile pump, tubing, a prosthetic heart valve, measurement sections and a reservoir. Peak systolic velocities (0.7-5.0 m/s, n = 10) were measured with standard MRI phase shift velocity mapping using different regions of interest (ROIs). Intravascular USD velocity measurements were carried out using a thin cannula measuring peak systolic local maximum and local mean velocity within the small sample volume.
Results: Using maximum velocity USD recordings, comparison with MRI velocities was poor when the ROI covered the complete vessel lumen giving offset -22.3% and limits of agreement (LOA) 57.5 and 105.1%. When the ROI was small the velocities were closer with offset -11.8% and LOA 75.3-103.3%. When the spatial resolution was comparable and the spatial mean was read from the USD recordings, the velocities showed good agreement with an offset close to zero (0.45%).
Conclusion: This study shows the importance of selecting small ROIs when using MRI for evaluating valve stenosis. Only when using spatially mean velocities over comparable regions the two methods showed good agreement.