Modeling Respiratory Movement Signals During Central and Obstructive Sleep Apnea Events Using Electrocardiogram

Obstructive sleep apnea (OSA) causes a pause in airflow with continuing breathing effort. In contrast, central sleep apnea (CSA) event is not accompanied with breathing effort. CSA is recognized when respiratory effort falls below 15% of pre-event peak-to-peak amplitude of the respiratory effort. The aim of this study is to investigate whether a combination of respiratory sinus arrhythmia (RSA), ECG-derived respiration (EDR) from R-wave amplitudes and wavelet-based features of ECG signals during OSA and CSA can act as surrogate of changes in thoracic movement signal measured by respiratory inductance plethysmography (RIP). Therefore, RIP and ECG signals during 250 pre-scored OSA and 150 pre-scored CSA events, and 10 s preceding the events were collected from 17 patients. RSA, EDR, and wavelet decomposition of ECG signals at level 9 (0.15–0.32 Hz) were used as input to the support vector regression (SVR) model to recognize the RIP signals and classify OSA from CSA. Using cross-validation test, an optimal SVR (radial basis function kernel; C = 2⁸ and ε = 2⁻² where C is the coefficient for trade-off between empirical and structural risk and ε is the width of ε-insensitive region) showed that it correctly recognized 243/250 OSA and 139/150 CSA events (95.5% detection accuracy). Independent test was performed on 80 OSA and 80 CSA events from 12 patients. The independent test accuracies of OSA and CSA detections were found to be 92.5 and 95.0%, respectively. Results suggest superior performance of SVR using ECG as the surrogate in recognizing the reduction of respiratory movement during OSA and CSA. Results also indicate that ECG-based SVR model could act as a potential surrogate signal of respiratory movement during sleep-disordered breathing.     

Electrocardiogram Derived Respiratory Rate from QRS Slopes and R-Wave Angle

A method for estimating respiratory rate from electrocardiogram (ECG) signals is presented. It is based on QRS slopes and R-wave angle, which reflect respiration-induced beat morphology variations. The 12 standard leads, 3 leads from vectorcardiogram (VCG), and 2 additional non-standard leads derived from VCG loops were analyzed. The following series were studied as ECG derived respiration (EDR) signals: slope between the peak of Q and R waves, slope between the peak of R and S waves, and the R-wave angle. Information from several EDR signals was combined in order to increase the robustness of estimation. Evaluation is performed over two databases containing ECG and respiratory signals simultaneously recorded during two clinical tests with different characteristics: tilt test, representing abrupt cardiovascular changes, and stress test representing a highly non-stationary and noisy environment. A combination of QRS slopes and R-wave angle series derived from VCG leads obtained a respiratory rate estimation relative error of 0.50 ± 4.11% (measuring 99.84% of the time) for tilt test and 0.52 ± 8.99% (measuring 96.09% of the time) for stress test. These results outperform those obtained by other reported methods, both in tilt and stress testing.     

Towards classifying non-segmented heart sound records using instantaneous frequency based features

Abstract

Heart sound and its recorded signal which is known as phonocardiograph (PCG) are one of the most important biosignals that can be used to diagnose cardiac diseases alongside electrocardiogram (ECG). Over the past few years, the use of PCG signals has become more widespread and researchers pay their attention to it and aim to provide an automated heart sound analysis and classification system that supports medical professionals in their decision. In this paper, a new method for heart sound features extraction for the classification of non-segmented signals using instantaneous frequency was proposed. The method has two major phases: the first phase is to estimate the instantaneous frequency of the recorded signal; the second phase is to extract a set of eleven features from the estimated instantaneous frequency. The method was tested into two different datasets, one for binary classification (Normal and Abnormal) and the other for multi-classification (Five Classes) to ensure the robustness of the extracted features. The overall accuracy, sensitivity, specificity, and precision for binary classification and multi-classification were all above 95% using both random forest and KNN classifiers.     

Separation of electrocardiographic and encephalographic components based on signal averaging and wavelet shrinkage techniques

During ambulatory monitoring, it is often required to record the electroencephalogram (EEG) and the electrocardiogram (ECG) simultaneously. It would be ideal if both EEG and ECG can be obtained with one measurement. We introduce an algorithm combining the wavelet shrinkage and signal averaging techniques to extract the EEG and ECG components from an EEG lead signal to a noncephalic reference (NCR). The evaluation using simulation data and measured data showed that the normalized power spectrum unvaried in all frequency bands for the EEG components, and the sensitivity and specificity of R-wave detection for the ECG component were nearly 100%.     

A robust approach for ECG-based analysis of cardiopulmonary coupling

Deriving respiratory signal from a surface electrocardiogram (ECG) measurement has advantage of simultaneously monitoring of cardiac and respiratory activities. ECG-based cardiopulmonary coupling (CPC) analysis estimated by heart period variability and ECG-derived respiration (EDR) shows promising applications in medical field. The aim of this paper is to provide a quantitative analysis of the ECG-based CPC, and further improve its performance. Two conventional strategies were tested to obtain EDR signal: R-S wave amplitude and area of the QRS complex. An adaptive filter was utilized to extract the common component of inter-beat interval (RRI) and EDR, generating enhanced versions of EDR signal. CPC is assessed through probing the nonlinear phase interactions between RRI series and respiratory signal. Respiratory oscillations presented in both RRI series and respiratory signals were extracted by ensemble empirical mode decomposition for coupling analysis via phase synchronization index. The results demonstrated that CPC estimated from conventional EDR series exhibits constant and proportional biases, while that estimated from enhanced EDR series is more reliable. Adaptive filtering can improve the accuracy of the ECG-based CPC estimation significantly and achieve robust CPC analysis. The improved ECG-based CPC estimation may provide additional prognostic information for both sleep medicine and autonomic function analysis.     

基于小波变换的R波检测算法

介绍了一种将小波变换应用于ECG信号检测R波的算法。该算法主要是在离散小波变换和多分辨率分析原理的基础上，利用db1小波特有的时频域特性，运用Mallat快速算法对心电信号进行了3层小波分解，分别在2、3层小波分解的高频系数中选择一定的阈值作为R波的判定条件，实现R波检测。通过对MIT／BIH（Massachusetts Institute of Technology／Boston’s Beth Israel Hospital）心电数据库的R波检测，结果表明，该检测算法即使在噪声干扰和病态的情况下，也很容易实现对R波的准确检测和精确定位，具有相当高的定位精度，R波正确检测率高达到99．8％。     

High-frequency oscillations of the heart rate during ramp load reflect the human anaerobic threshold

The aim of this study was to analyze the dynamic behavior of the high-frequency component (HF > 0.15 Hz) of heart rate variability (HRV) and the respiratory frequency in relation to the anaerobic threshold (AT). Twenty-two healthy subjects [mean (SD) age: 24 (6) years, height: 175 (10) cm, body mass: 65 (11) kg] completed a ramp load, with increments of 20 W · min⁻¹, on a cycle ergometer. The AT was determined by the V-slope-method. Respiratory movements of the thorax, and the electrocardiogram were monitored. The instantaneous frequency of the HF component of HRV and of the respiratory signal were obtained by the Hilbert transformation. Both frequencies were closely related, the cross correlation coefficient being between 0.84 and 0.99. Various patterns of HRV and respiration were observed during the protocol. Remarkably, however, in over 90% of these cases, a shift in the instantaneous frequency of the HF component occurred during the transition from aerobic to anaerobic work. The difference between the AT determined by gas analysis and the AT evaluated as the power output (AT _f ), calculated using the approximation of the curve of the instantaneous frequency of HF by hyperbolic tangent functions, varied between 2 and 14%. In conclusion, the present study demonstrates significant changes in the behavior of the instantaneous frequency of HF in the region of the AT. Accepted: 18 July 2000     

基于心电信号的脉搏波形特征点提取

为寻求一种精确的脉搏波特征提取方法,提取更多的脉搏波形特征,揭示心电脉搏在时域上的相关性,使用MP425数据采集卡和LabVIEW构成的数据采集系统同步采集ECG信号和脉搏波信号,对ECG信号和脉搏波信号进行分析和处理,采用能量算子法检测心电信号R波;基于同步采集的ECG和脉搏波信号,提出一种应用ECG信号的R波和T波来提取Pulse wave的重搏波和峰值的方法.经过分析与实验验证,该方法能准确找到脉搏波波峰和重搏波位置,并具有较强的抗干扰能力,为研究心电脉搏之间的关系提供了一种新的方法.     

Unconstrained detection of respiration rhythm and pulse rate with one under-pillow sensor during sleep

A completely non-invasive and unconstrained method is proposed to detect respiration rhythm and pulse rate during sleep. By employing wavelet transformation (WT), waveforms corresponding to the respiration rhythm and pulse rate can be extracted from a pulsatile pressure signal acquired by a pressure sensor under a pillow. The respiration rhythm was obtained by an upward zero-crossing point detection algorithm from the respiration-related waveform reconstructed from the WT 2⁶ scale approximation, and the pulse rate was estimated by a peak point detection algorithm from the pulse-related waveform reconstructed from the WT 2⁴ and 2⁵ scale details. The finger photo-electric plethysmogram (FPP) and nasal thermistor signals were recorded simultaneously as reference signals. The reference pulse rate and respiration rhythm were detected with the peak and upward zero-crossing point detection algorithm. This method was verified using about 24 h of data collected from 13 healthy subjects. The results showed that, compared with the reference data, the average error rates were 3.03% false negative and 1.47% false positive for pulse rate detection in the extracted pulse waveform. Similarly, 4.58% false negative and 3.07% false positive were obtained for respiration rhythm detection in the extracted respiration waveform. This study suggests that the proposed method is suitable, in sleep monitoring, for the diagnosis of sleep apnoea or sudden death syndrome.     

由单通道心电提取呼吸信息的算法

基于除呼吸作用以外，能够引起正常的QRS波群中R波振幅的干扰因素也较少的事实，提出使用室上性的QRS波群由心电提取呼吸信息（EDR）（ECG-derived respiratory signal）的算法。该算法先将单通道心电进行小波分解，再将小波近似在适当尺度上的室上性波R波振幅在时间上进行延拖，通过一低通滤波器后再将抽样频率降低至5Hz，再通过一带通滤波器（0．1～0．4Hz）后，即可得到EDR。     

一种基于多层感知器的房颤心电图检测方法

目的:提出一种基于多层感知器（MLP）的新型房颤识别算法。方法:首先设计一种新型自适应的R波阈值检测算法,然后以R波位置和幅度为特征,MLP为分类器进行正常/房颤心电图识别。MLP的网络参数采用深层置信网络预训练算法进行初始化,最后用误差反向传播算法对MLP网络权重进行调整。结果:在单通道心电图数据集上对正常、房颤心电信号进行分类,本研究方法的灵敏度达96.00%,特异性为84.18%,平均识别率为90.09%。结论:这种基于MLP的心电识别算法准确率高、计算复杂度较低,可为房颤的智能诊断提供一种新方法。     

Effect of ECG-derived respiration (EDR) on modeling ventricular repolarization dynamics in different physiological and psychological conditions

Ventricular repolarization dynamics is an important predictor of the outcome in cardiovascular diseases. Mathematical modeling of the heart rate variability (RR interval variability) and ventricular repolarization variability (QT interval variability) is one of the popular methods to understand the dynamics of ventricular repolarization. Although ECG derived respiration (EDR) was previously suggested as a surrogate of respiration, but the effect of respiratory movement on ventricular repolarization dynamics was not studied. In this study, the importance of considering the effect of respiration and the validity of using EDR as a surrogate of respiration for linear parametric modeling of ventricular repolarization variability is studied in two cases with different physiological and psychological conditions. In the first case study, we used 20 young and 20 old healthy subjects’ ECG and respiration data from Fantasia database at Physionet to analyze a bivariate QT–RR and a trivariate \({\text{QT}}{-}{\text{RR}}{-}{\text{RESP}}\,{\text{or}}\,{\text{QT}}{-}{\text{RR}}{-}{\text{EDR}}\) model structure to study the aging effect on cardiac repolarization variability. In the second study, we used 16 healthy subjects’ data from drivedb (stress detection for automobile drivers) database at Physionet to do the same analysis for different psychological condition (i.e., in stressed and no stress condition). The results of our study showed that model having respiratory information (QT–RR–RESP and QT–RR–EDR) gave significantly better fit value (p < 0.05) than that of found from the QT–RR model. EDR showed statistically similar (p > 0.05) performance as that of respiration as an exogenous model input in describing repolarization variability irrespective of age and different mental conditions. Another finding of our study is that both respiration and EDR-based models can significantly (p < 0.05) differentiate the ventricular repolarization dynamics between healthy subjects of different age groups and with different psychological conditions, whereas models without respiration or EDR cannot distinguish between the groups. These results established the importance of using respiration and the validity of using EDR as a surrogate of respiration in the absence of respiration signal recording in linear parametric modeling of ventricular repolarization variability in healthy subjects.     

Development of Gradient Descent Adaptive Algorithms to Remove Common Mode Artifact for Improvement of Cardiovascular Signal Quality

Background: Common-mode noise degrades cardiovascular signal quality and diminishes measurement accuracy. Filtering to remove noise components in the frequency domain often distorts the signal. Method: Two adaptive noise canceling (ANC) algorithms were tested to adjust weighted reference signals for optimal subtraction from a primary signal. Update of weight w was based upon the gradient term of the steepest descent equation: , where the error ɛ is the difference between primary and weighted reference signals. ∇ was estimated from Δɛ² and Δw without using a variable Δw in the denominator which can cause instability. The Parallel Comparison (PC) algorithm computed Δɛ² using fixed finite differences ± Δw in parallel during each discrete time k. The ALOPEX algorithm computed Δɛ²· Δw from time k to k + 1 to estimate ∇, with a random number added to account for Δɛ² · Δw→ 0 near the optimal weighting. Results: Using simulated data, both algorithms stably converged to the optimal weighting within 50–2000 discrete sample points k even with a SNR = 1:8 and weights which were initialized far from the optimal. Using a sharply pulsatile cardiac electrogram signal with added noise so that the SNR = 1:5, both algorithms exhibited stable convergence within 100 ms (100 sample points). Fourier spectral analysis revealed minimal distortion when comparing the signal without added noise to the ANC restored signal. Conclusions: ANC algorithms based upon difference calculations can rapidly and stably converge to the optimal weighting in simulated and real cardiovascular data. Signal quality is restored with minimal distortion, increasing the accuracy of biophysical measurement.     

Comparison of Respiratory Rates Derived from Heart Rate Variability, ECG Amplitude, and Nasal/Oral Airflow

It would often be desirable to obtain the respiratory rate during everyday conditions without obtaining an additional respiratory trace. This study investigates the agreement between respiratory rate assessed from the electrocardiogram (ECG) and the reference respiratory rate derived from a nasal/oral airflow (AF). Nasal/oral airflow and a Holter ECG were recorded in 52 healthy subjects (26 males, age range: 25.4–85.4 years) during everyday conditions for at least 10 h, including night-time sleep. The respiratory rate was assessed for each 5-min epoch (1) using respiratory sinus arrhythmia (RSA), (2) utilizing the respiration induced variations of the R-wave amplitude (ECG derived respiration (EDR)). The agreement with respect to AF was quantified using the average/std and the concordance correlation coefficient ρ_c. For RSA and EDR the difference with respect to AF was 0.2 cpm (std: 0.6 cpm) during sleep and −0.2 cpm (std: 1.0 cpm) during wake time. During sleep the RSA-approach performed best for subjects ≤50 years (ρ_c = 0.79) and worst for subjects >50 years (ρ_c = 0.41). The correlation of the EDR-approach was ρ_c = 0.73 for both groups. In conclusion, the respiratory rate may be assessed with reasonable agreement by both methods in younger subjects, but EDR should be preferred in the elderly.

Noise correlations and SNR in phased‐array MRS

The acquisition of magnetic resonance spectroscopy (MRS) signals by multiple receiver coils can improve the signal‐to‐noise ratio (SNR) or alternatively can reduce the scan time maintaining a reliable SNR. However, using phased array coils in MRS studies requires efficient data processing and data combination techniques in order to exploit the sensitivity improvement of the phased array coil acquisition method. This paper describes a novel method for the combination of MRS signals acquired by phased array coils, even in presence of correlated noise between the acquisition channels. In fact, although it has been shown that electric and magnetic coupling mechanisms produce correlated noise in the coils, previous algorithms developed for MRS data combination have ignored this effect. The proposed approach takes advantage of a noise decorrelation stage to maximize the SNR of the combined spectra. In particular Principal Component Analysis (PCA) was exploited to project the acquired spectra in a subspace where the noise vectors are orthogonal. In this subspace the SNR weighting method will provide the optimal overall SNR. Performance evaluation of the proposed method is carried out on simulated ¹H‐MRS signals and experimental results are obtained on phantom ¹H‐MR spectra using a commercially available 8‐element phased array coil. Noise correlations between elements were generally low due to the optimal coil design, leading to a fair SNR gain (about 0.5%) in the center of the field of view (FOV). A greater SNR improvement was found in the peripheral FOV regions. Copyright © 2009 John Wiley & Sons, Ltd.     

Time-frequency characterization of atrial fibrillation from surface ECG based on Hilbert-Huang transform

In this paper, based on Hilbert-Huang transform (HHT), we develop a new non-invasive time-frequency analysis method to characterize the dynamic behaviour of atrial fibrillation (AF) from surface ECG. We first extract f waves from single-lead ECG records of AF patients using PCA analysis. To capture the non-stationary behaviours of AF signals at different time scales, we use HHT to find the Hilbert spectrum and instantaneous frequency (IF) distribution of residual signals from principal component analysis. Two important feature variables, namely mean IF (mIF) and index of frequency stability over time (IS), are derived from the IF distribution, and in combination will be able to effectively discriminate two different AF types: self-terminating and non-terminating termination. The proposed AF signal decomposition and analysis method will help us efficiently differentiate individual AF patients, advance our understanding of AF mechanisms, and provide useful guidelines for improving administration of AF patients, especially paroxysmal AF.     

Validating a robust double‐quantum‐filtered 1H MRS lactate measurement method in high‐grade brain tumours

¹H MRS measurements of lactate are often confounded by overlapping lipid signals. Double‐quantum (DQ) filtering eliminates lipid signals and permits single‐shot measurements, which avoid subtraction artefacts in moving tissues. This study evaluated a single‐voxel‐localized DQ filtering method qualitatively and quantitatively for measuring lactate concentrations in the presence of lipid, using high‐grade brain tumours in which the results could be compared with standard acquisition as a reference. Paired standard acquisition and DQ‐filtered ¹H MR spectra were acquired at 3T from patients receiving treatment for glioblastoma, using fLASER (localization by adiabatic selective refocusing using frequency offset corrected inversion pulses) single‐voxel localization. Data were acquired from 2 × 2 × 2 cm³ voxels, with a repetition time of 1 s and 128 averages (standard acquisition) or 256 averages (DQ‐filtered acquisition), requiring 2.15 and 4.3 min respectively. Of 37 evaluated data pairs, 20 cases (54%) had measureable lactate (fitted Cramér–Rao lower bounds ≤ 20%) in either the DQ‐filtered or the standard acquisition spectra. The measured DQ‐filtered lactate signal was consistently downfield of lipid (1.33 ± 0.03 ppm vs 1.22 ± 0.08 ppm; p = 0.002), showing that it was not caused by lipid breakthrough, and that it matched the lactate signal seen in standard measurements (1.36 ± 0.02 ppm). In the absence of lipid, similar lactate concentrations were measured by the two methods (mean ratio DQ filtered/standard acquisition = 1.10 ± 0.21). In 7/20 cases with measurable lactate, signal was not measureable in the standard acquisition owing to lipid overlap but was quantified in the DQ‐filtered acquisition. Conversely, lactate was undetected in seven DQ‐filtered acquisitions but visible using the standard acquisition. In conclusion, the DQ filtering method has proven robust in eliminating lipid and permits uncontaminated measurement of lactate. This is important validation prior to use in tissues outside the brain, which contain large amounts of lipid and which are often susceptible to motion.     

Computerized quality assessment of phonocardiogram signal measurement-acquisition parameters

The major focus of this study is to describe and develop a phonocardiogram (PCG) signal measurement binary quality assessment (accept–reject) technique. The proposed algorithm is composed of three major stages: preprocessing, numerical-based quality measurement and advanced measurement subroutines. The preprocessing step includes normalization, wavelet-based threshold denoising and baseline wander removal. The numerical-based quality measurement routine includes two separate stages based on energy and level of noise of the PCG signal. The advanced quality measurement step is mainly based on the interval of S1 and S2 sounds. The proposed technique was applied to 400 2-min PCG signals gathered by volunteers with range of skills in PCG data acquisition from patients with different types of valve diseases from their 2R (aortic), 2L (pulmonic), 4R (apex) and 4L (tricuspid) positions by implementing an electronic stethoscope (3M Littmann^® 3200, 4 kHz sampling frequency). The dataset was firstly annotated manually and then, by applying the proposed algorithm, an accuracy of 95.25% was achieved.     

Retrospective motion gating in cardiac MRI using a simultaneously acquired navigator

A simultaneous acquisition technique of image and navigator signals (simultaneously acquired navigator, SIMNAV) is proposed for cardiac magnetic resonance imaging (CMRI) in Cartesian coordinates. To simultaneously acquire both image and navigator signals, a conventional balanced steady‐state free precession (bSSFP) pulse sequence is modified by adding a radiofrequency (RF) pulse, which excites a supplementary slice for the navigator signal. Alternating phases of the RF pulses make it easy to separate the simultaneously acquired magnetic resonance data into image and navigator signals. The navigator signals of the proposed SIMNAV were compared with those of current gating devices and self‐gating techniques for seven healthy subjects. In vivo experiments demonstrated that SIMNAV could provide cardiac cine images with sufficient image quality, similar to those from electrocardiogram (ECG) gating with breath‐hold. SIMNAV can be used to acquire a cardiac cine image without requiring an ECG device and breath‐hold, whilst maintaining feasible imaging time efficiency.