Heart signal recognition by Hidden Markov Models: the ECG case
Related Experiment Videos
L Thoraval1, G Carrault, J J Bellanger
1Laboratoire Traitement du Signal et de l'Image, Université de Rennes I, France.
Insights
Modified Continuous Variable Duration Hidden Markov Models (HMMs) improve ECG wave recognition by addressing limitations of stationary assumptions. This novel approach enhances performance compared to traditional HMMs in practice.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Machine Learning
Background:
- Traditional Hidden Markov Models (HMMs) for electrocardiogram (ECG) wave recognition assume stationary parameters, which is biologically unrealistic.
- This stationary assumption leads to significant errors in practical ECG signal analysis.
Purpose of the Study:
- To introduce a novel class of HMMs, Modified Continuous Variable Duration HMMs (MCVD-HMMs), specifically designed for ECG signal properties.
- To evaluate the efficacy of MCVD-HMMs when integrated with multiresolution front-end analysis for improved ECG recognition.
Main Methods:
- Development and application of Modified Continuous Variable Duration Hidden Markov Models (MCVD-HMMs).
- Integration of MCVD-HMMs with a multiresolution signal processing front-end for ECG analysis.
- Comparative performance evaluation against classical HMMs.
Main Results:
- The proposed MCVD-HMMs effectively account for the non-stationary characteristics of ECG signals.
- Coupling MCVD-HMMs with multiresolution analysis demonstrated enhanced ECG recognition capabilities.
- Significant performance improvements in ECG recognition were observed compared to conventional HMM approaches.
Conclusions:
- MCVD-HMMs offer a more natural and accurate modeling approach for ECG wave recognition.
- The proposed methodology represents a significant advancement over standard HMMs for clinical ECG analysis.
- This work highlights the potential of advanced HMMs for improving diagnostic accuracy in cardiology.
Abstract:
Wave recognition in ECG signals by Hidden Markov Models (HMMs) relies on the stationary assumption for the set of parameters used to describe ECG waves. This approach seems unnatural and consequently generates severe errors in practice. A new class of HMMs called Modified Continuous Variable Duration HMMs is proposed to account for the specific properties of the ECG signal. An application of the latter, coupled with a multiresolution front-end analysis of the ECG is presented. Results show these methods can increase the performance of ECG recognition compared to classical HMMs.