Algorithm for real-time detection of heart rate from noisy ECG signals supported by continuous blood pressure analysis

Abstract

The algorithm proposed in this paper is designed for robust identification of the heart beat annotations in multimodal data, consisting of ECG signal and one or several continuous arterial blood pressure signals. In case the ECG signal is distorted or unavailable the heart beat annotations are detected in continuous blood pressure signal. The novelty of the proposed solution lays in the adaptation of the algorithm for implementation on a real time system, a weighted estimation of the average RR interval in ECG signal and application of abnormality index estimation algorithm in advance to RR interval estimation from arterial blood pressure signal. The algorithm proposed in this paper reduced the HR estimation error from 6% to 1-2% for various SAI thresholds. For both analysed signal datasets the amount of F P annotations were reduced by our proposed algorithm, especially for the training dataset, where the amount of F P indications was reduced nearly 4 times.

Multiple Physiological Signals Fusion Techniques for Improving Heartbeat Detection: A Review

“…Table 6 presents the signals that have been employed in the reviewed papers. Most of the proposals for heartbeat detection employ the ECG and ABP signals, since these are directly related with cardiac activity [42,43,44,45,46,47,48,49,50,51,52]. Other authors have added some other signals to that group: PPG signal [53,54]; SV and PPG signals [55]; and EEG, EOG, and EMG signals [56,57].…”

“…ABP signal quality in References [47,50,55] and BP signal quality in Reference [68] are computed from the SAI, which integrates the pressure ranges and the average derivative for a cycle for the respective signals. The SAI is also employed to assess the ABP signal quality in References [42,46].…”

“…Usually, spurious annotations are directly removed whereas missing annotations are predicted using interpolation or the mean RR interval. This approach or similar ones are used in References [42,46,57,58,59,61,64,72,73,78,79]. The process may be repeated several times until convergence [81].…”

An empirical study of modified beat SQI based majority voting fusion method for heart-rate estimation in noisy multimodal cardiovascular signals

“…Several studies on heartbeat detection from the fusion of multimodal signals have been conducted, and overall reviews of such studies have been reported (Silva et al 2015). Based on the types of signals used for fusion, we can categorize previous studies in two ways: (i) the fusion of cardiovascular signals (Li et al 2007, Tarassenko et al 2010, Abromavicius and Serackis 2015; and (ii) the fusion of cardiovascular with NC signals (Gierałtowski et al 2015, Rankawat andDubey 2017). For studies exploring the fusion of cardiovascular signals, Li et al (2007Li et al ( , 2009 fused ECG with ABP signals, Tarassenko et al (2010) fused ECG with pulse oximetry signals, and Abromavicius and Serackis (2015) fused ECG signals with several continuous ABP leads.…”

“…Based on the types of signals used for fusion, we can categorize previous studies in two ways: (i) the fusion of cardiovascular signals (Li et al 2007, Tarassenko et al 2010, Abromavicius and Serackis 2015; and (ii) the fusion of cardiovascular with NC signals (Gierałtowski et al 2015, Rankawat andDubey 2017). For studies exploring the fusion of cardiovascular signals, Li et al (2007Li et al ( , 2009 fused ECG with ABP signals, Tarassenko et al (2010) fused ECG with pulse oximetry signals, and Abromavicius and Serackis (2015) fused ECG signals with several continuous ABP leads. However, these techniques would not give reliable heart-rate estimates if all the fused cardiovascular signals are concurrently noisy.…”