Abstract
Radiocardiography has been widely used as a method of quantification of cardiac output, by applying the principle of dye dilution method. Kuwahara and Hirakawa, with their co-worker, have reported successive studies of the analysis of radiocardiogram by analog computer simulation and its clinical applications. To analyze radiocardiogram by simulation is to estimate parameters of a model by curve fitting approach, and the analog computer acts as an effective tool in this parameter estimation. However, as the curve fitting technique requires a skilled operator of the analog computer, an automatic analyzing system of radiocardiogram using a minicomputer is expected to be developed especially for clinical use.
In this paper, we show a differential-difference equation model of transport process of radioisotope appearing in radiocardiogram, and discuss about an algorithm for a minicomputer to estimate the parameter values of the model in a reasonably short time. After Fourier transformation of two modified radiocardiograms, actual and modeled, a least square estimation is adopted in the frequency domain. Since the criterion function in this estimation does not change its basic expression form before and after Fourier transformation and the data to be compared are effectively compressed into the narrow frequency range, the computation time can be reduced fairly well. Parameter sensitivity analysis is applied to clarify the effects of parameters on the model and to shorten the computation time.
Some clinical results of radiocardiograms analyzed by thealgorithm are shown and it is verified that the algorithm developed here is sufficiently useful and efficient for clinical use.
