Structured sparsity models for compressively sensed electrocardiogram signals: A comparative study

Abstract

Abstract-We have recently quantified and validated the potential of the emerging compressed sensing (CS) paradigm for real-time energy-efficient electrocardiogram (ECG) compression on resource-constrained sensors. In the present work, we investigate applying sparsity models to exploit underlying structural information in recovery algorithms. More specifically, re-visiting well-known sparse recovery algorithms, we propose novel modelbased adaptations for the robust recovery of compressible signals like ECG. Our results show significant performance gains for the recovery algorithms exploiting the underlying sparsity models. I. INTRODUCTIONCS is a new sensing and processing paradigm, which challenges the traditional analog-to-digital conversion based on the Shannon sampling theorem. For sparse signals such as the electrocardiogram (ECG), Nyquist-rate sampling produces a large amount of redundant digital samples, which are costly to wirelessly transmit in the context of our target mobile ECG monitoring systems, and require to be further compressed using non-linear digital techniques. CS is a methodology that has been recently proposed to address this problem.Capitalizing on this sparsity, we have recently proposed [1], to apply the emerging compressed sensing (CS) approach [2] for a low-complexity, real-time and energy-efficient ECG signal compression on wireless body sensor network (WBSN) motes. We have also quantified the potential of the emerging compressed sensing (CS) signal acquisition/compression paradigm for low-complexity energy-efficient ECG compression on the state-of-the-art Shimmer TM WBSN mote. Interestingly, our results show that CS represents a competitive alternative to state-of-the-art digital wavelet transform (DWT)-based ECG compression solutions in the context of WBSNbased ECG monitoring systems. The results validates the suitability of compressed sensing for real-time energy-aware ECG compression on resource-constrained WBSN motes.Recent works about practical utilization of CS for biomedical signal analysis confirm that in order to be competitive with other state-of-the-art compression algorithms, new CS methods that fully leverage the underlying structural information of the considered signals (beyond simple sparsity analysis) must be developed. Building on our previous work, we herein investigate state-of-the-art CS recovery algorithms and apply model-based recovery for ECG signals.Notation: In all the following, normal letters designate scalar quantities, boldface lower-case letters indicate column

Soft computing based compressive sensing techniques in signal processing: A comprehensive review

“…Amongst them, power productivity can be improved through implanted ECG compression, to diminish broadcast appointment above energy-hungry remote connections. Hossein Mamaghanian et al [6], have evaluated the capability of the developing CS signal acquisition worldview for low-multifaceted nature energy effective ECG compression on the cutting edge Shimmer WBSN bit. Curiously, their outcomes demonstrate that CS speaks to an aggressive choice to cutting edge "Digital Wavelet Transform (DWT)" -based ECG compression arrangements.…”

Compressed sensing ECG using restricted Boltzmann machines

“…The application of CS to wireless ECG systems has attracted attention in recent years. Works in this area include studies of practical design considerations for CS encoding and decoding [3,6], as well as the development of CS reconstruction algorithms that exploit structural ECG information [7,8,9,10,11,12]. This paper proposes to use our recently developed CS scheme [13], henceforth referred to as RBM-CS scheme, for ECG reconstruction.…”

Power Line Interference Suppression for ECG Signal Recovery

“…Recently, in [5,6,12], the compressed sensing (CS) algorithm is applied to perform ECG compression, where basis pursuit (BP) and greedy algorithms including orthogonal matching pursuit (OMP), compressive sampling matching pursuit (CoSaMP), regularized orthogonal least squares and normalized iterative hard thresholding are utilized to perform the ECG recovery. By exploring transform domain, the CS concept has been successfully applied to perform more accurate reconstruction.…”