Finite rate Innovation and its applications in electrocardiography

Abstract

Compression and reconstruction of Electrocardiograms (ECG) has been a common topic of research in signal processing as wavelets, compressed sensing and many other methods have been used. However, the goal of this thesis was to demonstrate a sampling scheme, Variable Pulse Width Finite Rate of Innovation (VPW-FRI), whose samples provided meaningful information. The samples could be split into subspaces which were associated to the di fferent waveforms in an ECG and hence specfi c features could be reconstructed directly from the samples without having to reconstruct the entire signal. This allowed for many clinical applications such as heart rate determination, fetal heart rate calculation and P and T wave detection. This extended to a multichannel sampling scheme. Since this is a subspace method, it had robust de-noising capabilities. The novel subspace method developed in this thesis solves many of the earlier limitations of VPW-FRI and FRI and the e fficacy will be demonstrated on data from hospitals and medical devices.