Back to search

University of Southampton

Signal processing of electrocardiogram for arrhythmia prediction and classification

Abstract

dc:description.abstract

The prevalence of Cardiovascular Diseases (CVD) currently poses a significant global burden to the healthcare systems. Arrhythmia is a condition associated with CVD that can lead to death in chronic CVD patients or even to an apparent healthy person. Electrocardiogram (ECG) is the first-point diagnosis tool of arrhythmia. Currently, Implantable Cardioverter Defibrillator (ICD) and its subcutaneous version (S-ICD) are used for treating fatal arrhythmias. However, these devices can only detect arrhythmia once it started and therefore in the case of fatal arrhythmias such as Ventricular Fibrillation they are not very effective in saving lives. Recently, ECG has been applied to predict impending arrhythmias. However, their prediction and classification performance need to be improved to become acceptable in day-to-day clinical use. Also, such predictive methods need to be incorporated within the framework of ICD itself. This dissertation addresses this issue and provide a solution by improving the predictive capability of ECG and then integrating it with S-ICD. We first proposed two automated algorithms for precise delineation of ECG fiducial points based on the time- domain Hierarchical clustering and the time-frequency-domain Discrete Wavelet Transform (DWT). Then we developed a new algorithm based on the Phase Space Reconstruction (PSR) to overcome the T-wave over-sensing problem that typically plagues efficacy of ICDs, more specifically the S- ICD system. After that, we reported a risk index based on PSR and fuzzy c-means clustering to predict an impending fatal Ventricular Arrhythmias (VA) and its classifications (four different types of VA). The main feature of the proposed method is that for the first time it has been shown that it is not only possible to predict an impending arrhythmia sufficiently before its actual occurrence in time but also is possible to classify the type of arrhythmia before it actually occurs (1 min after the prediction time point). We further extended these methods for classifying other types of non-fatal arrhythmias such as, Atrial Fibrillation (AF) which are indicative to the risk of stroke in a patient suffering from chronic CVD condition. The methods proposed here could be used in day-to-day clinical practice after rigorous clinical trial to advance technologies such as ICD and S-ICD that can help to pre-empt the occurrence of fatal ventricular arrhythmia - a main cause of Sudden Cardiac Death (SCD).

Degree

thesis:*
Name dc:type.qualificationname
Ph.D.
Level dc:type.qualificationlevel
doctoral
Grantor dc:publisher.institution
University of Southampton
Year dc:date.issued
2022

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Chen, Hanjie
Advisor dc:contributor.advisor
  • Maharatna, Koushik
ORCID person:identifier.orcid

Rights

Language dc:language
English, eng

Chain of custody

source
Harvested from
University of Southampton
Base URL
eprints.soton.ac.uk/cgi/oai2
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Chen, Hanjie. Signal processing of electrocardiogram for arrhythmia prediction and classification. doctoral thesis, University of Southampton, 2022.