Assessing the performance of ultrafast vector flow imaging in the neonatal heart via multiphysics modeling and in-vitro experiments

Ultrafast vector flow imaging would benefit newborn patients with congenital heart disorders, but still requires thorough validation before translation to clinical practice. This study investigates 2D speckle tracking of intraventricular blood flow in neonates when transmitting diverging waves at ultrafast frame rate. Computational and in-vitro studies enabled us to quantify the performance and identify artefacts related to the flow and the imaging sequence. First, synthetic ultrasound images of a neonate's left ventricular flow pattern were obtained with the ultrasound simulator Field II by propagating point scatterers according to 3D intraventricular flow fields obtained with computational fluid dynamics (CFD). Non-compounded diverging waves (opening angle of 60°) were transmitted at a pulse repetition frequency of 9 kHz. Speckle tracking of the B-mode data provided 2D flow estimates at 180 Hz, which were compared to the CFD flow field. We demonstrated that the diastolic inflow jet showed a strong bias in the lateral velocity estimates at the edges of the jet, as confirmed by additional in-vitro tests on a jet flow phantom. Further, speckle tracking performance was highly dependent on the cardiac phase with low flows (< 5 cm/s), high spatial flow gradients and out-of-plane flow as deteriorating factors. Despite the observed artefacts, a good overall performance of 2D speckle tracking was obtained with a median magnitude underestimation and angular deviation of respectively 28% and 13.5° during systole, and 16% and 10.5° during diastole.