Flow quantification using fast cine phase-contrast MR imaging, conventional cine phase-contrast MR imaging, and Doppler sonography: in vitro and in vivo validation.

Abstract

Our purpose was to assess the accuracy of measurements of flow velocity and volume flow rate in an in vitro phantom and in healthy human volunteers using a cardiac-gated, segmented K-space, fast cine phase-contrast (PC) MR imaging technique with view sharing (fast PC). We compared this method with conventional cine PC MR imaging and Doppler sonography. Pulsatile flow was generated in a flow phantom that consisted of a cylindric tube having various degrees of tapered stenosis. Phase-encoded velocity maps were obtained using cine PC and fast PC MR imaging. Doppler sonography was also performed. Measurements of aortic and pulmonary artery peak systolic and minimum diastolic velocity and volume flow rate were then compared in eight healthy volunteers using the three imaging techniques. We found excellent agreement between fast PC and cine PC measurements of peak systolic velocity when regions of interest were drawn to exclude vessel margins (r > .99 for phantom studies, and r = .80 for human studies). Correlation between minimum diastolic velocity measurements by MR imaging was limited by noise that resulted from high encoding velocity settings. However, such correlation improved with signal averaging. When compared with predicted values of volume flow rates, both cine PC (r > .99) and fast PC (r = .97) MR imaging were more accurate than Doppler sonography (r = .78) in vitro. Measurements of cardiac output were adversely affected by low signal to noise, especially during diastole; estimates based on systolic forward flow resulted in better agreement between the two MR imaging methods. Fast PC MR flow quantification may prove to be a useful adjunct to routine MR studies for measurements of peak flow velocity. However, estimates of volume flow rate using fast PC MR imaging are limited because of increased noise during low diastolic flow as well as edge artifacts.