Indicator-dilution dispersion models and cardiac output computing methods.

Abstract

The general theory of indicator-dilution methods provides a basis for computing improved cardiac output estimates. Interpretation is via indicator-dispersion modeling with Brownian motion of drifting particles. Detected curves indicate the distribution of passage times from the injection site: the local density random walk (LDRW) function of a Wiener process. Fitting the LDRW to 70 dye curves by nonlinear regression for examples, I show how all possible undistorted curves can, in principle, be simulated. I show via semilogarithmic plots that conventional exponential decay constructs systematically underestimate cardiac output by up to 8%. To help reconcile the predictions of LDRW-fitted dilution curves and contemporary practice, I show how curve-shape asymmetry (skewness) dramatically affects the enclosed areas. Mean transit times may overestimate blood volumes by 15-100% in very skewed thermodilution curves if the dispersion effects are overlooked. Triangle constructions, which accounted for hundreds of experimental findings, also have theoretical explanations. Curve-fitting methods reduce the extrapolation biases inherent in many computers and in any respiration-induced artifacts. Compatibility of cardiac output predictions from various dilution methods and modules becomes feasible.