Deconvolution analysis of radionuclide angiocardiography curves: problems arising from fragmented bolus injections

Deconvolution analysis of radionuclide angiocardiography time-activity curves (TAC) in the cases of fragmented bolus injections was studied by using a new deconvolution technique based on the maximum a posteriori estimator. This method was first tested by simulation studies which showed its capability for accurate handling of deconvolution both for single and double peak input functions. These simulations have proved this technique to have advantages over a Fourier transform division method used with a low-pass filter. In patient studies, deconvolution of pulmonary TAC by superior vena caval TAC produced poor results in double peak (fragmented) bolus injections, compared with single peak bolus injections (SPBI). This discrepancy seems to be due to the fact that the postulates of linearity and invariance, on which deconvolution analysis is based, do not hold for the 'cardiopulmonary system' which extends from the superior vena cava to the lung. The differences between the model and the real system appear unimportant for SPBI. However, differences become apparent for fragmented bolus injections. Deconvolution analysis should be avoided with this type of input function in this cardiopulmonary system.