Commenges D, Brendel A J
Comput Programs Biomed. 1982 Jun;14(3):271-6. doi: 10.1016/0010-468x(82)90033-2.
We present a computer program for deconvolution of noisy signals. The unknown signal is modelled as a stochastic process and the solution is given by the 'maximum a posteriori estimator'. This is computed by two algorithms: the first uses and the Kalman filter; the second the conjugate gradient technique which permits introduction of a positivity constraint. The program has proved to give accurate results in stimulation studies. It has been applied to the deconvolution analysis of radionuclide angiocardiography time-activity curves. This improves the reliability of the results in left-to-right cardiac shunt quantitation.
