Estimating coronary blood flow using CT transluminal attenuation flow encoding: Formulation, preclinical validation, and clinical feasibility

Abstract

BackgroundWe present the formulation and testing of a new CT angiography (CTA)–based method for noninvasive measurement of absolute coronary blood flow (CBF) termed transluminal attenuation flow encoding (TAFE). CTA provides assessment of coronary plaque but does not allow for detection of vessel specific ischemia. A simple and direct method to calculate absolute CBF from a standard CTA could isolate the functional consequence of disease and aid therapy decisions. MethodsWe present the theoretical framework and initial testing of TAFE. Nine canine models of ischemic heart disease were prepared and underwent CT imaging and microsphere measurements of myocardial blood flow. Additionally, 39 acute chest pain patients with normal coronary arteries underwent coronary CTA. We applied TAFE to calculate absolute CBF in the coronary arteries using 4 vessel input parameters including transluminal attenuation gradient, cross-sectional area, length, and the contrast bolus duration derived from the arterial input function. ResultsIn animal studies, TAFE-derived CBF in the left anterior descending, left circumflex, and right coronary artery was 20.8 ± 10.4 mL/min, 27.0 ± 13.4 mL/min, and 6.0 ± 3.7 mL/min, respectively. TAFE-derived CBF divided by myocardial mass strongly correlated with microsphere myocardial blood flow (R2 = 0.90, P < .001). In human studies, TAFE-derived CBF in the left anterior descending, left circumflex, and right coronary artery was 26.4 ± 10.7 mL/min, 20.1 ± 13.0 mL/min, and 43.2 ± 40.9 mL/min, respectively. CBF per unit mass was 0.93 ± 0.48 mL/g/min in patients. Interobserver variability was minimal with excellent correlation (R = 0.96, P < .0001) and agreement (mean difference, 4.2 mL/min). ConclusionTAFE allows for quantification of absolute CBF from a standard CTA acquisition and may provide functional significance of coronary disease without complex computational methods.