Velocimetry based on multiphase CCTA images for reliable coronary flow velocity quantification across stenosis severities and ischemia detection.

BACKGROUND

Coronary flow velocity (CFV) serves as a critical functional feature for evaluating coronary stenosis severity (CSS) and detecting ischemia. With its noninvasive nature and rapid imaging, coronary CT angiography (CCTA) is expected to replace alternatives-computational fluid dynamics (CFD), magnetic resonance imaging (MRI), and digital subtraction angiography (DSA)-for hemodynamic evaluation and CFV quantification. Nevertheless, transluminal attenuation gradient-based velocimetry has been controversial, and its lack of dynamic flow information leads to appreciable quantified deviations. Although multiphase CCTA images within a single cardiac cycle inherently capture blood flow dynamics, a methodology leveraging this information for reliable CFV quantification has not yet been developed.

PURPOSE: This study aims to develop velocimetry based on multiphase CCTA images (VBMPI) for CFV quantification and to investigate the potential of CFV-derived hypothesis-generating indicators for characterizing CSS and providing ischemia detection.

METHODS

VBMPI quantifies CFV in multiple coronary ranges by tracking the transport process of the final maximum contrast agent concentration within a single cardiac cycle. First, CFV was quantified in a sophisticated 3D-printed in vitro right coronary artery (RCA) replica under five different heart rate conditions with the nonstenotic scenario to determine the optimal VBMPI parameter combination. Next, in vitro RCA replicas with five plaque configurations-each simulating CSS from 10% to 90%-were conducted by five independent trials under five heart rate conditions to evaluate the applicability and reproducibility of VBMPI. Quantification accuracy was validated against benchmark CFV obtained from in silico CFD simulations and in vitro MRI 4D flow scans. Based on quantified CFV in different ranges, multiple CFV-derived indicators were established via stratified CSS to investigate their exploratory utility in detecting myocardial ischemia. In addition, left coronary artery (LCA) replicas with the nonstenotic scenario were fabricated to verify the optimal VBMPI parameter combination and to perform in vitro assessments with varying stenotic scenarios, and representative patients were enrolled for an illustrative in vivo assessment of the overall methodology.

RESULTS

Within VBMPI, the combination of a 2 mm quantification plane spacing and a 1% delay phase increment was determined as the optimal parameter for accurate CFV quantification. Across all in vitro RCA replicas, VBMPI achieved a mean CFV deviation below 4% and a maximum deviation not exceeding 9%. Leveraging these reliable results, the ratio of CFV in the stenotic range to that in the proximal-mid range emerged as an exploratory indicator for CSS characterization and ischemia detection. Consistently, LCA replica assessments corroborated the selected parameters, with CFV quantification and CSS characterization aligning with benchmarks. In the five-patient illustrative cohort, stenotic range CFV was quantified as 0.600, 0.700, 0.600, 0.723, and 0.815 m/s, whereas proximal-mid range CFV was 0.426, 0.469, 0.273, 0.221, and 0.335 m/s. The estimated CSS was consistent with the DSA findings, and the estimated ischemia was qualitatively consistent with deep-learning-derived FFR.

CONCLUSION

As a preliminary methodology, VBMPI shows prospective promise for image-based CFV quantification, offering functional insights beyond anatomical evaluation, which may complement existing CCTA protocols to strengthen noninvasive ischemia detection.