Abstract
Remote telemonitoring of physiological parameters is being used increasingly both for medical research and for clinical management. However, the technique entails the collection of large volumes of information for transmission and storage which necessitates efficient means of data compression. This paper presents a simple mathematical model of the lossy compression of physiological signals. The model is developed in a top-down style making design decisions progressively. First a general model of lossy compression is developed and some lemmas presented. Then the model is refined to take advantage of inherent cyclicity in many types of physiological signal. Finally the model is specialized to the compression of electrocardiograms. An algorithm is described for implementing the model. The algorithm is a table-based method capable of achieving very low output bit rates (50 bps) with reasonable fidelity. This compares favourably with the results obtained by other workers and holds promise for further evaluation. The generality of the model should ultimately allow a coherent approach to the compression of the various types of physiological signal currently being recorded.