A fast signal quality assessment algorithm applied to a contactless electrocardiogram system

Abstract

Signal quality assessment (SQA), or noise detector systems, promise to improve the quality of the collected electrocardiogram (ECG). This is achieved by enabling automated systems to accurately collect and process ECG data, or facilitating telemetry through the accurate selection of acceptable data, in order to be analyzed remotely by a cardiologist. This thesis is a research & development (R&D) collaboration with SIG.NUM, a company that is specialized in the development and commercialization of an innovative proprietary contactless and automated technology for both short and long term ECG acquisition, known as cECG. This work presents fundamentals regarding different ECG instrumentation systems; a review about SQA, its fundaments and applications to cECG systems; a review about wavelets and its applications regarding ECG signal representation; and finally, the design of an SQA classifier system, or noise detector, using artificial neural networks (ANN), and wavelet scalogram-based fast signal processing techniques for feature extraction purposes. The SQA classifier system is meant to be used as a real-time channel selector for the cECG system developed by SIG.NUM. The developed machine learning based algorithm for ECG SQA is trained by a merge of several annotated data-sets, available from PhysioNet (Goldberger et al., 2000). Results of the designed system are mainly compared with the single-lead SQA technique proposed by the work of Clifford et al. (2012). In conclusion, this thesis showed that a simple shallow ANN is sufficient for well modelling a generalized SQA classifier system. Moreover, results showed that the wavelet transform feature extraction method, proposed by this thesis, is not only very powerful to represent ECG signal quality, but also computationally efficient.

Date	24 Mar 2021
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Ghyslain Gagnon (Supervisor)