A SIMULATION-BASED SENSITIVITY STUDY OF RADIONUCLIDE ANGIOCARDIOGRAPHIC APPROACHES TO SHUNT ASSESSMENT

There are currently 2 popular approaches to RNA [radionuclide angiocardiography] shunt assessment, the Stewart-Hamilton method and the .gamma.-variate method. The goodness of these methods is generally evaluated by regression analysis against O2-saturation data. This method of evaluation is at best uncertain (which variable is independent?) and at worst misleading. A computer-simulation model for evaluating the accuracy of RNA shunt assessment techniques is described. The model is based on a directed-graph representation of the central circulation and continuous stochastic flow of tracer. Sensitivity analyses varying just the injection form for the RNA study and the transit time of the shunt loop indicate that RNA shunt assessment cannot yet be considered exact. Substantial variation, up to 10 or 20% of cardiac output, could easily result from repeated measurements on the same patient.