P and T wave analysis in ECG signals using Bayesian methods

Abstract

This thesis studies Bayesian estimation/detection algorithms for P and T wave analysis in ECG signals. In this work, different statistical models and associated Bayesian methods are proposed to solve simultaneously the P and T wave delineation task (determination of the positions of the peaks and boundaries of the individual waves) and the waveform-estimation problem. These models take into account appropriate prior distributions for the unknown parameters (wave locations and amplitudes, and waveform coefficients). These prior distributions are combined with the likelihood of the observed data to provide the posterior distribution of the unknown parameters. Due to the complexity of the resulting posterior distributions, Markov chain Monte Carlo algorithms are proposed for (sample-based) detection/estimation. On the other hand, to take full advantage of the sequential nature of the ECG, a dynamic model is proposed under a similar Bayesian framework. Sequential Monte Carlo methods (SMC) are also considered for delineation and waveform estimation. In the last part of the thesis, two Bayesian models introduced in this thesis are adapted to address a specific clinical research problem referred to as T wave alternans (TWA) detection. One of the proposed approaches has served as an efficient analysis tool in the Endocardial T wave Alternans Study (ETWAS) project in collaboration with St. Jude Medical, Inc and Toulouse Rangueil Hospital. This project was devoted to prospectively assess the feasibility of TWA detection in repolarisation on EGM stored in ICD memories.