Comparison of four mathematical models to analyze indicator-dilution curves in the coronary circulation

While several models have proven to result in accurate estimations when measuring cardiac output using indicator dilution, the mono-exponential model has primarily been chosen for deriving coronary blood/plasma volume. In this study, we compared four models to derive coronary plasma volume using indicator dilution; the mono-exponential, power-law, gamma-variate, and local density random walk (LDRW) model. In anesthetized goats (N = 14), we determined the distribution volume of high molecular weight (2,000 kDa) dextrans. A bolus injection (1.0 ml, 0.65 mg/ml) was given intracoronary and coronary venous blood samples were taken every 0.5-1.0 s; outflow curves were analyzed using the four aforementioned models. Measurements were done at baseline and during adenosine infusion. Absolute coronary plasma volume estimates varied by ~25% between models, while the relative volume increase during adenosine infusion was similar for all models. The gamma-variate, LDRW, and mono-exponential model resulted in volumes corresponding with literature, whereas the power-model seemed to overestimate the coronary plasma volume. The gamma-variate and LDRW model appear to be suitable alternative models to the mono-exponential model to analyze coronary indicator-dilution curves, particularly since these models are minimally influenced by outliers and do not depend on data of the descending slope of the curve only.