[Paper Review] A tool for ECG signal analysis using standard and optimized Hermite transform

Zoja Vulaj, Andjela Draganić+2 more|arXiv (Cornell University)|Mar 1, 2017

Blind Source Separation Techniques5 references4 citations

TL;DR

This paper presents a Matlab-based tool for ECG signal analysis using standard and optimized Hermite transforms, leveraging the shape similarity between QRS complexes and Hermite basis functions to achieve high signal concentration with fewer coefficients. The optimized parametrized Hermite transform reduces data representation requirements, outperforming Fourier transform in energy compaction and enabling efficient signal transmission and classification for clinical and portable monitoring applications.

ABSTRACT

The development of a system that would ease the diagnosis of heart diseases would also fasten the work of the cardiologic department in hospitals and facilitate the monitoring of patients with portable devices. This paper presents a tool for ECG signal analysis which is designed in Matlab. The Hermite transform domain is exploited for the analysis. The proposed transform domain is very convenient for ECG signal analysis and classification. Parts of the ECG signals, i.e. QRS complexes, show shape similarity with the Hermite basis functions, which is one of the reasons for choosing this domain. Also, the information about the signal can be represented using a small set of coefficients in this domain, which makes data transmission and analysis faster. The signal concentration in the Hermite domain and consequently, the number of samples required for signal representation, can additionally be reduced by performing the parametization of the Hermite transform. For the comparison purpose, the Fourier transform domain is also implemented within the software, in order to compare the signal concentration in two transform domains.

Motivation & Objective

• To develop a computationally efficient ECG signal analysis tool for clinical and portable monitoring applications.
• To exploit the Hermite transform domain for its ability to represent ECG signals with fewer coefficients due to shape similarity with QRS complexes.
• To compare the performance of standard and parametrized Hermite transforms against the Fourier transform in terms of signal concentration and data reduction.
• To enable faster data transmission and processing by minimizing the number of significant coefficients in the transform domain.

Proposed method

• The ECG signal is transformed into the Hermite transform domain using standard Hermite basis functions.
• Parametrization of the Hermite transform is applied to optimize signal concentration by adjusting parameters to match QRS complex characteristics.
• The Fourier transform is implemented as a comparative benchmark within the same Matlab environment.
• Signal energy concentration is evaluated by analyzing the number of significant coefficients required for accurate signal reconstruction.
• The tool is designed to support real-time analysis and data transmission, with emphasis on low computational load.
• The system is validated using ECG signals, comparing coefficient sparsity and reconstruction fidelity across transform domains.

Experimental results

Research questions

• RQ1 How well does the Hermite transform domain represent ECG signals compared to the Fourier transform in terms of energy compaction?
• RQ2 To what extent can parametrization of the Hermite transform reduce the number of required coefficients for signal representation?
• RQ3 Can the Hermite transform domain effectively capture the morphological features of QRS complexes due to shape similarity with basis functions?
• RQ4 What is the trade-off between computational complexity and signal representation efficiency in the Hermite domain versus the Fourier domain?

Key findings

• The Hermite transform domain achieves higher signal concentration than the Fourier transform, requiring fewer coefficients for accurate ECG signal representation.
• Parametrization of the Hermite transform further reduces the number of significant coefficients, enhancing data compression efficiency.
• The shape similarity between QRS complexes and Hermite basis functions enables effective signal representation with minimal information loss.
• The optimized Hermite transform outperforms the standard Fourier transform in energy compaction, supporting faster and more efficient signal transmission.