Baseline Drift Tolerant Signal Encoding for ECG Classification with Deep Learning

Common artefacts such as baseline drift, rescaling, and noise critically limit the performance of machine learning-based automated ECG analysis and interpretation. This study proposes Derived Peak (DP) encoding, a non-parametric method that generates signed spikes corresponding to zero crossings of the signal’s first and second-order time derivatives. Notably, DP encoding is invariant to shift and scaling artefacts, and its implementation is further simplified by the absence of user-defined parameters. DP encoding was used to encode the 12-lead ECG data from the PTB-XL dataset (n=18,869 participants) and was fed to 1D-ResNet-18 models trained to identify myocardial infarction, conductive deficits and ST-segment abnormalities. Robustness to artefacts was assessed by corrupting ECG data with sinusoidal baseline drift, shift, rescaling and noise, before encoding. The addition of these artefacts resulted in a significant drop in accuracy for seven other methods from prior art, while DP encoding maintained a baseline AUC of 0.88 under drift, shift and rescaling. DP achieved superior performance to unencoded inputs in the presence of shift (AUC under 1 mV shift: 0.91 vs 0.62), and rescaling artefacts (AUC 0.91 vs 0.79). Thus, DP encoding is a simple method by which robustness to common ECG artefacts may be improved for automated ECG analysis and interpretation.

I. Introduction

Deep neural networks have demonstrated promising results for automated electrocardiograph (ECG) classification and monitoring [1]. However, intensive energy requirements have been identified as a potential barrier to clinical implementation within the resource constraints of implanted and wearable devices [1]. Event-driven ECG processing limits communication to salient events such as waveform complexes, thereby reducing energy requirements [2] –[6]. "Spike"-based encoding optimises efficiency and latency further through ECG representation as sparse sequences of "all-or-none" spike signals. Spiking neural network models (SNNs) employ spike encoding throughout, facilitating accurate ECG classification on microjoule-range energy budgets [6]. Thus, SNNs present an opportunity to both improve battery-life and response latency in implanted cardiac defibrillators and pacemakers and wearable monitors [2] –[5]. However, ECG signals are subject to significant distributional variability under normal operating conditions [7], [8], presenting a major reliability obstacle for spike-based cardiovascular monitoring and diagnosis. Physiological factors such as respiration and limb movement cause baseline drift, a low-frequency oscillatory artefact [9]. In contrast, diagnostic ECG features are subtle, as a 0.2 mV ST segment elevation may signify myocardial infarction [10]. This problem is considered one of the most significant challenges for robust ECG classification [1], [7], [8], [11]. Event-driven ECG encoding techniques typically convert analogue input to spikes according to fine-tuned threshold criteria [4], [6], [12]. Consequently, these methods are particularly vulnerable to commonly encountered artefacts such as baseline drift [11], [13]. To improve the robustness of spike-based ECG encoding to these factors, we propose a parameter-free method based on zero-crossings of first and second-order time derivatives.