A dynamic nonlinear time domain model for reconstruction and compression of cardiovascular signals with application to telemedicine.

A new nonlinear time domain model is proposed in this paper for signals of cardiovascular origin. An equation of the dynamic nonlinear model has been obtained by considering a masking function, which is modulated by a harmonic series with the baseline drift incorporated into the model. Signal reconstruction using model parameters has established the effectiveness of the model for signal compression. Improvement has been effected by using neural networks for reducing the time for optimizing the initial parameters. An improved adaptive optimization step size algorithm has also been implemented. Results show that the technique is able to provide reasonable compression with low error between the original and reconstructed signals. One of the main advantages of the model is its potential of being used for compression of many different types of biosignals transmitted in parallel. Incorporation of the compression model into a telemedicine system has led to considerable saving in transmission time for patient data.