Beat Template Research Articles

Statistical shape modeling-based algorithm for replacing missing beats in blood pressure signals.

In clinical environments, such as the Intensive Care Unit (ICU), continuous and uninterrupted monitoring of vital signs is critical for the early detection of patient deterioration and prompt intervention. Since data collected in these settings are often corrupted by noise, artifacts, or recording gaps, it is important to estimate missing data for a more accurate signal assessment.In this study, we propose an automatic algorithm for reconstructing of arterial blood pressure signal waveforms. The methodological core of the algorithm is based on the idea of statistical shape modeling, which basically estimates the shape variation of beat waveforms in order to reconstruct them in noisy segments. The waveform reconstruction is achieved by combining the average beat template from a 90-second segment of clean signal preceding the gap with the main shape variations of the estimated waveform.The algorithm was validated using arterial blood pressure recordings from 9 subjects admitted in the ICU and collected in the MIMIC-III Waveform Database, each lasting 1 hour and sampled at 125 Hz. For each record, ten fictitious gaps were created, and the reconstructed segments were compared to the original signals with the metrics proposed within the PhysioNet / Computing in Cardiology Challenge 2010. Results demonstrate the excellent performance of the proposed algorithm, with overall averages of both Q1 and Q2 metrics greater than 0.85.

M-beat-A novel marker for selective left bundle branch capture.

Premature-ventricular-complexes (template/fixation beat) guided left bundle branch pacing (LBBP) was recently described as a novel method of successful lead deployment by rapid rotations. We aimed at analyzing the incidence of a unique morphology template beat, which we labelled as 'M-beat' in patients undergoing PVC-guided LBBP, its ability to predict selective LBB-capture and clinical significance. Overall 210 out of 217 attempted-patients (96.7%) underwent successful LBBP. Template beat was noted in 90.4% patients (n = 190) and M-beat in 32.8%(n = 69). Non-selective to selective capture transition demonstrated in 55.2%(n = 116). The QRS duration of the M-beat was 129.3 ± 13.1ms. Patients were divided into two groups: Group-I with M-beat (n = 69;32.8%) and Group-II without M-beat (n = 141; 67.2%). The mean fluoroscopy-time was significantly less in group-I as compared to group-II (13.1 ± 11.1 vs 16.8 ± 12.04 minutes; p-0.03). Patients in group-II required more attempts as compared to group-I for successful lead deployment (2.8 ± 1.09 vs 2.2 ± 1.04; p - 0.01). Six patients showed loss of R-wave in lead-V1 and 2 showed rise in LBB capture threshold by >1V during follow-up in group-II. M-beat had a specificity of 96.77% and sensitivity of 58.62% (positive-predictive-value-98.55%) to predict selective-LBB capture. Myocardial excitability would not modify the occurrence of M-beat as opposed to capture transition response since it could be demonstrated without pacing protocols. When confirmation of LBB-capture itself would be difficult in patients with baseline LBBB-morphology, M-beat with 42.8% incidence predicted selective capture with 96.7% specificity and 66.04% sensitivity(positive-predictive-value-97.22%). M-beat is a marker of transient-selective LBB-capture, independent of the local myocardial excitability with high specificity and positive predictive value irrespective of the baseline QRS morphology.

PPG Signal Morphology-Based Method for Distinguishing Stress and Non-Stress Conditions

In this study, the morphology of the PPG signal has been analyzed to be a potential cardiovascular marker for physiological stress. The morphology of the PPG signal was quantified as signal quality index by comparing the template beat (extracted from resting conditions) to the PPG beats recorded during vigorous physical activity. Data was taken from eight subjects where they performed some physical activities ranging from low to high intensity. It was found that, the mean and standard deviation of correlation coefficient between non-stress condition template beat and annotated PPG beat, 89.43±5.17 (%) and 44.23±10.48 (%) for non-stress and stress beat respectively with P value of 2.04*10-06 shows significantly difference between correlation coefficients (stress and non-stress). Whereas, mean and standard deviation of dynamic time warping correlation coefficients are 93.43±5.06 (%) and 85.93±4.18 (%) for non-stress and stress beat respectively with P value of .04. The morphology results corroborate the findings from the traditional HRV parameters generally used for stratifying stress.

MoDTRAP: Improved heart rate tracking and preprocessing of motion-corrupted photoplethysmographic data for personalized healthcare

ObjectiveA new joint framework for heart rate (HR) tracking and preprocessing of motion artifact (MA) corrupted photoplethysmogram (PPG) data is presented. MethodsTwo popular signal decomposition techniques, viz., variational mode decomposition (VMD) to track HR, and ensemble empirical mode decomposition (EEMD) in conjunction with an artificial neural network (NN) model was utilized for preprocessing. The HR frequency was estimated from the signal strengths of VMD decomposed mode functions and tracked with an indigenous algorithm utilizing the fact that the PPG signal normally comprises base HR frequency along with its harmonics. For MA reduction, The HR synchronous signal component was extracted from the appropriate mode functions of VMD decomposition and reconstructed to minimize MA using a template matching technique utilizing an autoencoder and pre-trained multilayer feedforward neural network model. A combination of EEMD followed by a LSTM binary classifier was used to generate and update the reference PPG template beat for the above purpose. ResultsWith 2015 IEEE signal processing cup challenge database, an average absolute error and percent absolute error of 1.02 and 0.86 respectively compared to the ground truth HR were obtained for all 22 subjects. For MA reduction, we used wrist PPG data collected from 30 human subjects (normal and cardiovascular patients), achieving an average root mean squared error (RMSE) of 0.31 and SNR improvement of 21.23 dB. ConclusionProposed MoDTRAP technique provides noticeable improvements, separately, over existing methods on HR tracking and MA reduction. SignificanceThe proposed technique can be utilized for ambulatory healthcare monitoring.

ECG Multilead QT Interval Estimation Using Support Vector Machines.

This work reports a multilead QT interval measurement algorithm for a high-resolution digital electrocardiograph. The software enables off-line ECG processing including QRS detection as well as an accurate multilead QT interval detection algorithm using support vector machines (SVMs). Two fiducial points (Qini and Tend) are estimated using the SVM algorithm on each incoming beat. This enables segmentation of the current beat for obtaining the P, QRS, and T waves. The QT interval is estimated by updating the QT interval on each lead, considering shifting techniques with respect to a valid beat template. The validation of the QT interval measurement algorithm is attained using the Physionet PTB diagnostic ECG database showing a percent error of 2.60 ± 2.25 msec with respect to the database annotations. The usefulness of this software tool is also tested by considering the analysis of the ECG signals for a group of 60 patients acquired using our digital electrocardiograph. In this case, the validation is performed by comparing the estimated QT interval with respect to the estimation obtained using the Cardiosoft software providing a percent error of 2.49 ± 1.99 msec.

Improving Photoplethysmographic Measurements Under Motion Artifacts Using Artificial Neural Network for Personal Healthcare

Photoplethysmographic (PPG) measurements are susceptible to motion artifacts (MA) due to movement of the peripheral body parts. In this paper, we present a new approach to identify the MA corrupted PPG beats and then rectify the beat morphology using artificial neural network (ANN). Initially, beat quality assessment was done to identify the clean PPG beats by a pre-trained feedback ANN to generate a reference beat template for each person. The PPG data was decomposed using principal component analysis (PCA) and reconstructed using fixed energy retention. A weight coefficient was assigned for each PPG samples in such a way that when they are multiplied , the modified beat morphology matches the reference template. A particle swarm optimization (PSO) based technique was utilized to select the best weight weight vector coefficients to tune another feedback ANN, fed with a set of significant features generated by an auto encoder from PCA reconstructed data. For real time implementation, this pre-trained ANN was operated in feed-forward mode to directly generate the weight vectors for any subsequent measurements of PPG. The method was validated with PPG data collected from 55 human subjects. An average RMSE of 0.28 and SNR improvement of 14.54 dB was obtained, with an average improvement of 36% and 47% measurement accuracy on crest time and systolic to diastolic peak height ratio respectively. With IEEE Signal Processing Cup 2015 Challenge database, Pearson's correlation coefficient between PPG estimated and ECG derived heart rate was 0.990. The proposed method can be useful for personal health monitoring applications.

Influence of automatic frequent pace-timing adjustments on effective left ventricular pacing during cardiac resynchronization therapy.

Cardiac resynchronization therapy (CRT) requires effective left ventricular (LV) pacing (i.e. sufficient energy and appropriate timing to capture). The AdaptivCRT™ (aCRT) algorithm serves to maintain ventricular fusion during LV or biventricular pacing. This function was tested by comparing the morphological consistency of ventricular depolarizations and percentage effective LV pacing in CRT patients randomized to aCRT vs. echo-optimization. Continuous recordings (≥20 h) of unipolar LV electrograms from aCRT (n = 38) and echo-optimized patients (n = 22) were analysed. Morphological consistency was determined by the correlation coefficient between each beat and a template beat. Effective LV pacing of paced beats was assessed by algorithmic analysis of negative initial EGM deflection in each evoked response. The %CRT pacing delivered, %effective LV pacing (i.e. % of paced beats with effective LV pacing), and overall %effective CRT (i.e. product of %CRT pacing and %effective LV pacing) were compared between aCRT and echo-optimized patients. Demographics were similar between groups. The mean correlation coefficient between individual beats and template was greater for aCRT (0.96 ± 0.03 vs. 0.91 ± 0.13, P = 0.07). Although %CRT pacing was similar for aCRT and echo-optimized (median 97.4 vs. 98.6%, P = 0.14), %effective LV pacing was larger for aCRT [99.6%, (99.1%, 99.9%) vs. 94.3%, (24.3%, 99.8%), P=0.03]. For aCRT vs. echo-optimized groups, the proportions of patients with ≥90% effective LV pacing was 92 vs. 55% (P = 0.002), and with ≥90% effective CRT was 79 vs. 45%, respectively (P = 0.018). AdaptivCRT™ significantly increased effective LV pacing over echo-optimized CRT.

Superiority of Classification Tree versus Cluster, Fuzzy and Discriminant Models in a Heartbeat Classification System.

This study presents a 2-stage heartbeat classifier of supraventricular (SVB) and ventricular (VB) beats. Stage 1 makes computationally-efficient classification of SVB-beats, using simple correlation threshold criterion for finding close match with a predominant normal (reference) beat template. The non-matched beats are next subjected to measurement of 20 basic features, tracking the beat and reference template morphology and RR-variability for subsequent refined classification in SVB or VB-class by Stage 2. Four linear classifiers are compared: cluster, fuzzy, linear discriminant analysis (LDA) and classification tree (CT), all subjected to iterative training for selection of the optimal feature space among extended 210-sized set, embodying interactive second-order effects between 20 independent features. The optimization process minimizes at equal weight the false positives in SVB-class and false negatives in VB-class. The training with European ST-T, AHA, MIT-BIH Supraventricular Arrhythmia databases found the best performance settings of all classification models: Cluster (30 features), Fuzzy (72 features), LDA (142 coefficients), CT (221 decision nodes) with top-3 best scored features: normalized current RR-interval, higher/lower frequency content ratio, beat-to-template correlation. Unbiased test-validation with MIT-BIH Arrhythmia database rates the classifiers in descending order of their specificity for SVB-class: CT (99.9%), LDA (99.6%), Cluster (99.5%), Fuzzy (99.4%); sensitivity for ventricular ectopic beats as part from VB-class (commonly reported in published beat-classification studies): CT (96.7%), Fuzzy (94.4%), LDA (94.2%), Cluster (92.4%); positive predictivity: CT (99.2%), Cluster (93.6%), LDA (93.0%), Fuzzy (92.4%). CT has superior accuracy by 0.3–6.8% points, with the advantage for easy model complexity configuration by pruning the tree consisted of easy interpretable ‘if-then’ rules.

Wavelet Entropy as a Measure of Ventricular Beat Suppression from the Electrocardiogram in Atrial Fibrillation

A novel method of quantifying the effectiveness of the suppression of ventricular activity from electrocardiograms (ECGs) in atrial fibrillation is proposed. The temporal distribution of the energy of wavelet coefficients is quantified by wavelet entropy at each ventricular beat. More effective ventricular activity suppression yields increased entropies at scales dominated by the ventricular and atrial components of the ECG. Two studies are undertaken to demonstrate the efficacy of the method: first, using synthesised ECGs with controlled levels of residual ventricular activity, and second, using patient recordings with ventricular activity suppressed by an average beat template subtraction algorithm. In both cases wavelet entropy is shown to be a good measure of the effectiveness of ventricular beat suppression.

Oral Abstract Session: Novel non-invasive risk marker

Oral Abstract Session: Novel non-invasive risk marker

072 NOVEL NON INVASIVE DETECTION OF ARRHYTHMIA SUBSTRATE USING SUPPORT VECTOR MACHINE LEARNING

Introduction Myocardial scar burden, quantified by MRI, predicts ventricular arrhythmogenesis. MRI use is resource and cost limited in contrast to conventional 12 lead ECG. Manual assessment of ECG time domain features lacks specificity for scar. We hypothesise that frequency and phase domain ECG analysis could yield data indicative of myocardial scar burden. A supervised learning method (support vector machine, SVM) was used to develop a novel computerised algorithm capable of screening ECGs for the presence of myocardial scar, so allowing large data volumes to be processed. Methods 153 consecutive adult patients (age 63±12, male 65%), attending for cardiac MRI with scar analysis for clinical reasons were recruited, and underwent digital ECG acquisition (500Hz, 18bit) at the time of scanning. Semi automated quantification of late gadolinium contrast enhancement was used to identify myocardial scar. The ECGs from those with no scar were used to construct a median template beat for each of the standard 12 leads. ECGs from 35 patients with scar and 35 with no scar were used to train the SVM by feature comparison with the median beat. The algorithm was then tested on the remaining 83 ECGs. Results 16 comparative parameters were included: cross correlation, covariance, magnitude and phase of wavelet coherence, mean value, median value, variance, SD, inter-quartile range, skewness, kurtosis, mobility and complexity of the power spectrum, Hurst exponent slope, detrended fluctuation factor, and differential entropy. Clinical and ECG characteristics in those with scar (age 64±11, LVEF 56±17%, QRS width 115±22 ms) and no scar (age 61±13, LVEF 60±19, QRS width 110±18 ms) were similar (p=NS). The test set consisted of ECGs from 72 patients with scar and 11 patients with no scar. SVM was able to classify the test ECGs with 80.6% sensitivity and 72.7% specificity. Conclusions Digital ECG analysis of frequency and phase signals, using a newly developed SVM, characterises features which classify myocardial scar, but our observations require confirmation in larger prospective studies. The sensitivity and specificity of this approach could make it valuable in population screening for more costly complex investigations such as MRI scanning.

Amplitude differences least squares method applied to temporal cardiac beat alignment

High resolution averaged ECG is an important diagnostic technique in post-infarcted and/or chagasic patients with high risk of ventricular tachycardia (VT). It calls for precise determination of the synchronism point (fiducial point) in each beat to be averaged. Cross-correlation (CC) between each detected beat and a reference beat is, by and large, the standard alignment procedure. However, the fiducial point determination is not precise in records contaminated with high levels of noise. Herein, we propose an alignment procedure based on the least squares calculation of the amplitude differences (LSAD) between the ECG samples and a reference or template beat. Both techniques, CC and LSAD, were tested in high resolution ECG's corrupted with white noise and 50 Hz line interference of varying amplitudes (RMS range: 0-100μV). Results point out that LSDA produced a lower alignment error in all contaminated records while in those blurred by power line interference better results were found only within the 0–40 μV range. It is concluded that the proposed method represents a valid alignment alternative.

Enhanced Real-Time ECG Coder for Packetized Telecardiology Applications

A new real-time compression method for electrocardiogram (ECG) signals has been developed based on the wavelet transform approach. The method is specifically adaptable for packetized telecardiology applications. The signal is segmented into beats and a beat template is subtracted from them, producing a residual signal. Beat templates and residual signals are coded with a wavelet expansion. Compression is achieved by selecting a subset of wavelet coefficients. The number of selected coefficients depends on a threshold which has different definitions depending on the operational mode of the coder. Compression performance has been tested using a subset of ECG records from MIT-BIH Arrhythmia database. This method has been designed for real-time packetized telecardiology scenarios both in wired and wireless environments.

A COMPARISON OF SIMILARITY MEASURES FOR CLUSTERING OF QRS COMPLEXES

Similarity or distance measures play important role in the performance of algorithms for ECG clustering problems. This paper compares four similarity measures such as the city block (L1-norm), Euclidean (L2-norm), normalized correlation coefficient, and simplified grey relational grade for clustering of QRS complexes. Performances of the measures include classification accuracy, threshold value selection, noise robustness, execution time, and the capability of automated selection of templates. The clustering algorithm used is the so-called two-step unsupervised method. The best out of the 10 independent runs of the clustering algorithm with randomly selected initial template beat for each run is used to compare the performances of each similarity measure. To investigate the capability of automated selection of templates for ECG classification algorithms, we use the cluster centers generated by the clustering algorithm with various measures as templates. Four sets of templates are obtained, each set for a measure. And the four sets of templates are used in the k-nearest neighbor classification method to evaluate the performance of the templates. Tested with MIT/BIH arrhythmia data, we observe that the simplified grey relational grade outperforms the other measures in classification accuracy, threshold value selection, noise robustness, and the capability of automated selection of templates.

Fast and reliable QRS alignment technique for high-frequency analysis of signal-averaged ECG.

The process of QRS alignment as required in signal-averaged ECG can impose serious limitations on the spectral range of the signal output. This effect depends basically on the particular alignment technique being used and on the level and type of noise present in the recorded ECG. In clinical studies where a wide-band (1000 Hz) ECG averager is required, the conventional QRS alignment technique, based on maximum coherence matching (MCM) with a template beat, may not perform consistently well. An alternative QRS alignment technique based on the accurate detection of a single fiducial point (SFP) in the bandpass filtered (3-30 Hz) QRS complex was developed. Using computer simulation methods, a comparative assessment of the frequency bandwidths (3 dB points) offered by both MCM and SFP techniques as a function of noise level (15-100 muRMS) and type (EMG and 50 Hz interference), was carried out. The results of the comparative assessment indicated a better performance by the SFP technique in all cases of noise. Hence, the SFP technique would perform more reliably for high-frequency analysis of a noisy ECG, especially when 50 Hz interference is high. Furthermore, SFP is considerably faster than MCM (about four times) when implemented digitally, and its analogue realisation is feasible. The SFP technique is suitable for late-potential analysis in the signal-averaged ECG.

Morphologic Differences of the Endocardial Electrogram in Beats of Sinus and Ventricular Origin*

The lack of accurate arrhythmia detection and identification is one of the major obstacles to improvement in the efficacy of antitachycardia devices. We evaluated a method for detection of beats of ventricular origin compared to sinus rhythm based on the morphology of the endocardial electrogram. In order to compare mechanically induced ventricular beats to normal sinus beats, endocardial electrograms from a standard pacing electrode were recorded from eight open-chested dogs. Time and frequency domain features analyzed included peak-to-peak amplitude (AMP), maximal slew rate (dV/dT), and frequency content (-3 dB downpoint). Quantitative morphologic comparison of the waveforms was performed using standard correlation and by the absolute area of difference between the waveform and a sinus beat template. The AMP and dV/dT for a group of ventricular beats did not differ significantly from beats of sinus origin. In the unipolar configuration -3 dB for ventricular beats was significantly different from sinus beats (p = .01), but overlap occurred in three of eight cases. Conversely, using either method of assessment of morphological differences, all ventricular beats could be identified without overlapping the values for normal beats. We concluded that morphologic analysis of the endocardial electrogram by such methods may be a highly accurate means of distinguishing between beats of sinus and ventricular origins. This technique may also be applicable to the problem of automatic rhythm identification by implanted devices.

Clinical evaluation of an improved high resolution ECG cart for recording the His bundle electrogram non-invasively.

The Marquette high resolution MCA-1 ECG cart can record the His bundle electrogram non-invasively at the bedside. The second generation MAC-1 has several unique features and technical advances which may now make the technique clinically applicable. In addition to amplification and signal averaging, an adjustable input filtration system excludes beats which are unacceptably "noisy" compared to a 4 beat template. Thus, extremely "clean" signals are presented for processing by the average allowing resolution of very low amplitude signals. Clear unequivocal His bundle recordings were obtainable in 25 of 55 patients (45%). A correlation coefficient of +.97 was obtained between invasive and non-invasive HV intervals. The 55% failure rate was attributed to prolonged atrial activity in 31%, short PR in 9%, and technically unsatisfactory recordings with excessive or overwhelming noise which did not allow appreciable signal resolution in 15%.