Exploiting similar prior knowledge for compressing ECG signals

https://doi.org/10.1016/j.bspc.2020.101960Get rights and content

Highlights

  • A simple peak detection technique is proposed for ECG signal.

  • Aligning detected beats is used to cancel the hidden redundancy between successive frames.

  • A selective compression technique is proposed to compress a redundancy removed similar frame or exact transmission of a dissimilar frame.

  • Performance comparison with the existing ECG-CS approaches are presented based on CR, PRD, Se, Psim, SP, FDM and energy consumption.

Abstract

Background and objectives

Data compression techniques have been used in order to reduce power consumption when transmitting electrocardiogram (ECG) signals in wireless body area networks (WBAN). Among these techniques, compressed sensing allows sparse or compressible signals to be encoded with only a small number of measurements. Although ECG signals are not sparse, they can be made sparse in another domain. Numerous sparsifying techniques are available, but when signal quality and energy consumption are important, existing techniques leave room for improvements.

Methods

To leverage compressed sensing, we increased the sparsity of an ECG frame by removing the redundancy in a normal frame. In this study, by framing a signal according to the detected QRS complex (R peaks), consecutive frames of the signal become highly similar. This helps remove redundancy and consequently makes each frame sparse. In order to increase detection performance, different frames that symptomize a cardiovascular disease are sent uncompressed.

Results

For evaluating and comparing our proposed technique with different state-of-the-art techniques two datasets that contained normal and abnormal ECG: MIT-BIH Arrhythmia Database and MIT-BIH Long Term Database were used. For performance evaluation, we performed heart rate variability (HRV) analysis as well as energy-based distortion analysis. The proposed method reaches an accuracy of 99.9%, for a compression ratio of 25. For MIT-BIH Long Term Database, the average percentage root-mean squared difference (PRD) is less than 10 for all compression ratios.

Conclusion

Removing the redundancy between successive similar frames and exact transmission of dissimilar frames, the proposed method proves to be appropriate for heart rate variability analysis and abnormality detection.

Introduction

A wireless body area network (WBAN) connects independent nodes (e.g. sensors and actuators) that are placed on the body or under the skin of a person. The network typically expands over the whole human body and the nodes are connected through a wireless communication channel [1]. WBANs serve a variety of applications like sports, defense, emotion detection, personal health monitoring, posture detection, medical consumer electronics [2]. Personal health monitoring is an ongoing trend, and the latest consumer electronic products confirm that. It is a developing method in personalized medicine and home-based e-health, enabling real-time monitoring of biological signals, among which electrocardiogram (ECG) signals for heart activity monitoring. ECG signals are used to detect heart rate variability (HRV), the position of normal and abnormal heartbeats, which can be caused by damage to heart muscle cells, the damage to the conduction system, the effects of cardiac drugs or the effects of an implanted pacemaker. The sensors are equipped with a limited amount of power supply and in most cases, replacing or recharging the battery is difficult, so energy consumption presents a challenge for the extensive use of WBANs. Sensors consume energy for sensing, processing and most for data transmission. To reduce the impact of data transmission on energy consumption, various data compression techniques can be implemented, but using such a technique comes at the expense of signal quality.

The study in this paper focuses on using similarity between ECG pulses because they tend to be similar to each other [3]. To the best of our knowledge, this is the first time high likeness between consecutive ECG frames is exploited for sparsifying frames prior to compressed sensing (CS). The proposed method can be used to detect abnormalities reliably because the abnormal frames that could be signs of heart diseases are not similar to normal frames. When abnormal frames are detected, they are sent intact.

The proposed redundancy removal-based method was applied to several ECG records at several compression ratios to check the capability of this technique for detecting HRV. The Pan-Tompkins [4] detection algorithm was also applied to the whole reconstructed signal for performance assessment. Experimental results show that the proposed technique provides little increase in the sensors’ processing time compared to plain CS, with the advantage of very good performance over all compression ratios (CR) and small errors for CRs above 3.5 compared to other techniques.

The main contribution of the paper is the data compression scheme that exploits the likeness between consecutive ECG frames for frame sparsification prior to compressed sensing and determines on the fly whether a frame is abnormal. While redundancy is removed between similar frames for sparsification, abnormal frames are sent without using CS technique, to ensure relevant signal characteristics are preserved for medical diagnosis without over-consuming energy.

In the rest of this paper, after a brief review of related work in Section 2, a description of ECG signals and compressed sensing is given in Section 3. Section 4 describes the proposed scheme. Section 5 presents an evaluation of the proposed work and compares it with several state-of-the-art techniques.

Section snippets

Related work

The energy consumption of physiological sensors, particularly ECG sensors, has always been a challenging issue, so a variety of lossy and lossless techniques that reduce energy expended in the transmission of ECG frames have been introduced. Lossless compression techniques, mostly extract static redundancies existence in the signal to reduce the total bit length. Codebook-based approaches are a popular format of these techniques, where values according to their frequency of occurrence are

Background

We begin this section by providing a brief introduction to ECG. We then cover some important aspects of the theory of compressed sensing. Throughout this paper, we use bold lower-case letters to denote a vector e.g. x, and bold upper-case letters for matrices, e.g., X. Scalar values are denoted by italic symbols, such as x. Additionally, xˆ denotes the estimated/recovered value of vector x. Table 1 defines some frequently used symbols in this paper.

Proposed scheme

The aim of this research is to improve energy-efficiency while maintaining the detection accuracy of HRV and heart problems. The workflow of the proposed scheme is summarized in Fig. 4. The workflow of the proposed technique starts with segmenting the digitized signal into 512 sample, enough to contain at least one QRS complex. After applying a simple R peak-detection on the segment, frame of aligned R-peak with previous R-peak is provided. According to the similarity between current frame and

Simulation and results

In order to validate the proposed technique, clinical ECG signals were used. These signals contained baseline wander and motion artifacts for accurate performance evaluation.

Conclusions

Compressed sensing theory is an attractive solution for WBAN, as it lowers power consumption in sensors by reducing the data size. Due to non-sparse sensed signals in WBAN, this area has not gained much benefit. In order to use compressed sensing the signal should be sparse or made sparse by transforming it to another domain using a sparsifying matrix. Most of the natural signals such as ECG are not sparse. So this research is aimed at using compressed sensing for these signals by removing the

Author contributions

Fahimeh Nasimi: Conceptualization, methodology, software, validation, investigation, writing – original draft. Mohammad Reza Khayyambashi: Resources, writing – review & editing, supervision. Naser Movahhedinia: Conceptualization, validation, formal analysis, writing – review & editing. Yee Wei Law: Conceptualization, validation, writing – review & editing.

Conflict of interest

The authors declare that there is no conflict of interest.

References (46)

  • T.A. Welch

    A technique for high-performance data compression

    Computer

    (1984)
  • K. Li et al.

    Real-time lossless ECG compression for low-power wearable medical devices based on adaptive region prediction

    Electron. Lett.

    (2014)
  • S.-L. Chen et al.

    Efficient fuzzy-controlled and hybrid entropy coding strategy lossless ECG encoder VLSI design for wireless body sensor networks

    Electron. Lett.

    (2013)
  • C.J. Deepu et al.

    An ECG-on-chip with 535 nW/channel integrated lossless data compressor for wireless sensors

    IEEE J. Solid-State Circuits

    (2014)
  • Z. Arnavut

    ECG signal compression based on burrows-wheeler transformation and inversion ranks of linear prediction

    IEEE Trans. Biomed. Eng.

    (2007)
  • D.A. Dipersio et al.

    Evaluation of the fan method of adaptive sampling on human electrocardiograms

    Med. Biol. Eng. Comput.

    (1985)
  • B.A. Rajoub

    An efficient coding algorithm for the compression of ECG signals using the wavelet transform

    IEEE Trans. Biomed. Eng.

    (2002)
  • R. Benzid et al.

    Fixed percentage of wavelet coefficients to be zeroed for ECG compression

    Electron. Lett.

    (2003)
  • M. Pooyan et al.

    Wavelet compression of ECG signals using SPIHT algorithm

    Int. J. Signal Process.

    (2004)
  • H. Imai et al.

    An efficient encoding method for electrocardiography using spline functions

    Syst. Comput. Japan

    (1985)
  • D.L. Donoho

    Compressed sensing

    IEEE Trans. Inf. Theory

    (2006)
  • H. Mamaghanian et al.

    Compressed sensing for real-time energy-efficient ECG compression on wireless body sensor nodes

    IEEE Trans. Biomed. Eng.

    (2011)
  • K. Xu et al.

    An energy-efficient compressive sensing framework incorporating online dictionary learning for long-term wireless health monitoring

  • View full text