A two-stage discrimination of cardiac arrhythmias using a total least squares-based prony modeling algorithm

Prony Analysis of Left Ventricle Pressure and Volume

Direct measurement of cardiac pressure-volume (PV) relationships is the gold standard for assessment of ventricular hemodynamics, but few innovations have been made to "multi-beat" PV analysis beyond traditional signal processing. The Prony method solves the signal recovery problem with a series of dampened exponentials or sinusoids. It achieves this by extracting the amplitude, frequency, dampening, and phase of each component. Since its inception, application of the Prony method to biologic and medical signal has demonstrated a relative degree of success, as a series of dampened complex sinusoids easily generalizes to multifaceted physiological processes. In cardiovascular physiology, the Prony analysis has been used to determine fatal arrythmia from electrocardiogram signals. However, application of the Prony method to simple left ventricular function based on pressure and volume analysis is absent. We have developed a new pipeline for analysis of pressure volume signals recorded from the left ventricle. We propose fitting pressure-volume data from cardiac catheterization to the Prony method for pole extraction and quantification of the transfer function. We implemented the Prony algorithm using open-source Python packages and analyzed the pressure and volume signals before and after severe hemorrhagic shock, and after resuscitation with stored blood. Each animal (n = 6 per group) underwent a 50% hemorrhage to induce hypovolemic shock, which was maintained for 30 min, and resuscitated with 3-week-old stored RBCs until 90% baseline blood pressure was achieved. Pressure-volume catheterization data used for Prony analysis were 1 s in length, sampled at 1000 Hz, and acquired at the time of hypovolemic shock, 15 and 30 min after induction of hypovolemic shock, and 10, 30, and 60 min after volume resuscitation. We next assessed the complex poles from both pressure and volume waveforms. To quantify deviation from the unit circle, which represents deviation from a Fourier series, we counted the number of poles at least 0.2 radial units away from it. We found a significant decrease in the number of poles after shock (p = 0.0072 vs. baseline) and after resuscitation (p = 0.0091 vs. baseline). No differences were observed in this metric pre and post volume resuscitation (p = 0.2956). We next found a composite transfer function using the Prony fits between the pressure and volume waveforms and found differences in both the magnitude and phase Bode plots at baseline, during shock, and after resuscitation. In summary, our implementation of the Prony analysis shows meaningful physiologic differences after shock and resuscitation and allows for future applications to broader physiological and pathophysiological conditions.

Coding Prony's method in MATLAB and applying it to biomedical signal filtering

BACKGROUND

The response of many biomedical systems can be modelled using a linear combination of damped exponential functions. The approximation parameters, based on equally spaced samples, can be obtained using Prony's method and its variants (e.g. the matrix pencil method). This paper provides a tutorial on the main polynomial Prony and matrix pencil methods and their implementation in MATLAB and analyses how they perform with synthetic and multifocal visual-evoked potential (mfVEP) signals. This paper briefly describes the theoretical basis of four polynomial Prony approximation methods: classic, least squares (LS), total least squares (TLS) and matrix pencil method (MPM). In each of these cases, implementation uses general MATLAB functions. The features of the various options are tested by approximating a set of synthetic mathematical functions and evaluating filtering performance in the Prony domain when applied to mfVEP signals to improve diagnosis of patients with multiple sclerosis (MS).

RESULTS

The code implemented does not achieve 100%-correct signal approximation and, of the methods tested, LS and MPM perform best. When filtering mfVEP records in the Prony domain, the value of the area under the receiver-operating-characteristic (ROC) curve is 0.7055 compared with 0.6538 obtained with the usual filtering method used for this type of signal (discrete Fourier transform low-pass filter with a cut-off frequency of 35 Hz).

CONCLUSIONS

This paper reviews Prony's method in relation to signal filtering and approximation, provides the MATLAB code needed to implement the classic, LS, TLS and MPM methods, and tests their performance in biomedical signal filtering and function approximation. It emphasizes the importance of improving the computational methods used to implement the various methods described above.

Filtering multifocal VEP signals using Prony's method

BACKGROUND

This paper describes use of Prony's method as a filter applied to multifocal visual-evoked-potential (mfVEP) signals. Prony's method can be viewed as an extension of Fourier analysis that allows a signal to be decomposed into a linear combination of functions with different amplitudes, damping factors, frequencies and phase angles.

METHOD

By selecting Prony method parameters, a frequency filter has been developed which improves signal-to-noise ratio (SNR). Three different criteria were applied to data recorded from control subjects to produce three separate datasets: unfiltered raw data, data filtered using the traditional method (fast Fourier transform: FFT), and data filtered using Prony's method.

RESULTS

Filtering using Prony's method improved the signals' original SNR by 44.52%, while the FFT filter improved the SNR by 33.56%. The extent to which signal can be separated from noise was analysed using receiver-operating-characteristic (ROC) curves. The area under the curve (AUC) was greater in the signals filtered using Prony's method than in the original signals or in those filtered using the FFT.

CONCLUSION

filtering using Prony's method improves the quality of mfVEP signal pre-processing when compared with the original signals, or with those filtered using the FFT.

Electrocardiogram pattern recognition and analysis based on artificial neural networks and support vector machines: a review

Computer systems for Electrocardiogram (ECG) analysis support the clinician in tedious tasks (e.g., Holter ECG monitored in Intensive Care Units) or in prompt detection of dangerous events (e.g., ventricular fibrillation). Together with clinical applications (arrhythmia detection and heart rate variability analysis), ECG is currently being investigated in biometrics (human identification), an emerging area receiving increasing attention. Methodologies for clinical applications can have both differences and similarities with respect to biometrics. This paper reviews methods of ECG processing from a pattern recognition perspective. In particular, we focus on features commonly used for heartbeat classification. Considering the vast literature in the field and the limited space of this review, we dedicated a detailed discussion only to a few classifiers (Artificial Neural Networks and Support Vector Machines) because of their popularity; however, other techniques such as Hidden Markov Models and Kalman Filtering will be also mentioned.

ARTIFICIAL NEURAL NETWORK BASED ECG ARRHYTHMIA CLASSIFICATION

Reliable and computationally efficient means of classifying electrocardiogram (ECG) signals has been the subject of considerable research effort in recent years. This paper explores the potential applications of a talented, versatile computation model called the Artificial Neural Network (ANN) in the field of ECG signal classification. Two types of ANNs: Multi-Layered Feed Forward Network (MLFFN) and Probabilistic Neural Networks (PNN) are used to classify seven types of ECG beats. It includes six types of arrhythmia data and normal data. Here, parametric modeling strategies are used in conjunction with ANN classifiers to discriminate ECG signals. Instead of giving the ECG data as such, parameters such as fourth order Auto Regressive model coefficients and Spectral Entropy of the signals has been selected. On testing with the Massachusetts Institute of Technology-Beth Israel Hospital (MIT/BIH) arrhythmia database, it has been observed that PNN has better performance than conventionally used MLFFN in ECG arrhythmia classification. MLFFN with Back Propagation Algorithm gives a classification accuracy of 97.54% and PNN gives 98.96%. The classification by PNN also has an advantage that the computation time for classification is lower than that of MLFFN.

Classification of arrhythmia using hybrid networks

Reliable detection of arrhythmias based on digital processing of Electrocardiogram (ECG) signals is vital in providing suitable and timely treatment to a cardiac patient. Due to corruption of ECG signals with multiple frequency noise and presence of multiple arrhythmic events in a cardiac rhythm, computerized interpretation of abnormal ECG rhythms is a challenging task. This paper focuses a Fuzzy C- Mean (FCM) clustered Probabilistic Neural Network (PNN) and Multi Layered Feed Forward Network (MLFFN) for the discrimination of eight types of ECG beats. Parameters such as fourth order Auto Regressive (AR) coefficients along with Spectral Entropy (SE) are extracted from each ECG beat and feature reduction has been carried out using FCM clustering. The cluster centers form the input of neural network classifiers. The extensive analysis of Massachusetts Institute of Technology- Beth Israel Hospital (MIT-BIH) arrhythmia database shows that FCM clustered PNNs is superior in cardiac arrhythmia classification than FCM clustered MLFFN with an overall accuracy of 99.05%, 97.14%, respectively.

Electrocardiogram signals identification for cardiac arrhythmias using prony's method and neural network

A new method is presented to identify Electrocardiogram (ECG) signals for abnormal heartbeats based on Prony's modeling algorithm and neural network. Hence, the ECG signals can be written as a finite sum of exponential depending on poles. Neural network is used to identify the ECG signal from the calculated poles. Algorithm classification including a multi-layer feed forward neural network using back propagation is proposed as a classifying model to categorize the beats into one of five types including normal sinus rhythm (NSR), ventricular couplet (VC), ventricular tachycardia (VT), ventricular bigeminy (VB), and ventricular fibrillation (VF).

Fuzzy clustered probabilistic and multi layered feed forward neural networks for electrocardiogram arrhythmia classification

The role of electrocardiogram (ECG) as a noninvasive technique for detecting and diagnosing cardiac problems cannot be overemphasized. This paper introduces a fuzzy C-mean (FCM) clustered probabilistic neural network (PNN) for the discrimination of eight types of ECG beats. The performance has been compared with FCM clustered multi layered feed forward network (MLFFN) trained with back propagation algorithm. Important parameters are extracted from each ECG beat and feature reduction has been carried out using FCM clustering. The cluster centers form the input of neural network classifiers. The extensive analysis using the MIT-BIH arrhythmia database has shown an average classification accuracy of 97.54% with FCM clustered MLFFN and 99.58% with FCM clustered PNN. Fuzzy clustering improves the classification speed as well. The result reveals the capability of the FCM clustered PNN in the computer-aided diagnosis of ECG abnormalities.

OURSES: a telemedicine project for rural areas in france. Telemonitoring of elderly people

Several telemedicine applications are proposed within the frame of OURSES project, French acronym for Offer of Rural Use of Services by Satellite, providing services for elderly people. The main objective of this project is to show the interest of using satellites as a complement to terrestrial technologies, in areas where telecommunication infrastructure is lacking or incomplete. This paper describes one of these applications: an ECG monitoring system. This telemonitoring system allows, thanks to a wireless wearable sensor, to detect possible cardiac problems of elderly people. ECG signals are analyzed through signal processing algorithms and if some abnormal condition is detected, an alarm is raised and sent via satellite to the physician's office. The corresponding physician is able to access at any time the recorded ECG signals, whenever he is willing to, in the presence of an alarm or not. This allows a constant monitoring of the elderly people. Tests realized in a real environment have demonstrated the feasibility and the interest of this application.

Model estimation of cerebral hemodynamics between blood flow and volume changes: a data-based modeling approach

It is well known that there is a dynamic relationship between cerebral blood flow (CBF) and cerebral blood volume (CBV). With increasing applications of functional MRI, where the blood oxygen-level-dependent signals are recorded, the understanding and accurate modeling of the hemodynamic relationship between CBF and CBV becomes increasingly important. This study presents an empirical and data-based modeling framework for model identification from CBF and CBV experimental data. It is shown that the relationship between the changes in CBF and CBV can be described using a parsimonious autoregressive with exogenous input model structure. It is observed that neither the ordinary least-squares (LS) method nor the classical total least-squares (TLS) method can produce accurate estimates from the original noisy CBF and CBV data. A regularized total least-squares (RTLS) method is thus introduced and extended to solve such an error-in-the-variables problem. Quantitative results show that the RTLS method works very well on the noisy CBF and CBV data. Finally, a combination of RTLS with a filtering method can lead to a parsimonious but very effective model that can characterize the relationship between the changes in CBF and CBV.

Multifractal analysis of ventricular fibrillation and ventricular tachycardia

A study of ventricular fibrillation and ventricular tachycardia was undertaken using multifractal analysis. By applying the method of direct determination of the f(alpha) singularity spectrum, the value of the area of the VF and VT singularity spectrum was calculated. The comparison between the results showed that the value of the area of the VF singularity spectrum tended to be larger than that of the value of the area of the VT singularity spectrum. This makes the multifractal singularity spectrum a powerful criterion for discriminating between VF and VT.

Subtraction of 16.67 Hz railroad net interference from the electrocardiogram: application for automatic external defibrillators

The widespread application of automatic external defibrillators (AEDs) for treating out-of-hospital cardiac arrest incidents and their particular use at railway stations defines the task for 16.67 Hz power line interference elimination from the electrocardiogram (ECG). Although this problem exists only in five European countries, it has to be solved in all AEDs, which must comply with the European standard for medical equipment requirements for interchangeability and compatibility between rail systems. The elimination of the railroad interference requires a specific approach, since its frequency band overlaps with a significant part of the frequencies in the QRS spectra. We present a method based only on one channel ECG signal processing, which effectively subtracts the interference components. The computation procedure is based on simple signal processing tools, which include: (i) bi-directional band-pass filtering (13-23 Hz) of the analyzed ECG segment; (ii) estimation of adequate linearity thresholds; (iii) frequency measurement and calculation of dynamic interference buffer in linear segments and (iv) phase synchronization and subtraction technique in nonlinear segments. The developed method has proved advantageous in providing sufficient quality of the output interference free ECG signal for seven defined arrhythmia types (normal sinus rhythm, normal rhythm, supraventricular tachicardia, slow/rapid ventricular tachycardia, and coarse/fine ventricular fibrillation), and simulated interferences with constant or variable frequencies and amplitudes, which cover the entire amplitude range of the input channel. The procedure is suitable to be embedded in AEDs as a preprocessing branch, which proves reliable for analysis of ECG signals, thus guaranteeing the specified accuracy of the AED automatic rhythm analysis algorithms.

Assessment of ECG frequency and morphology parameters for automatic classification of life-threatening cardiac arrhythmias

The reliable recognition and adequate electrical shock therapy of life-threatening cardiac states depend on the electrocardiogram (ECG) descriptors which are used by the defibrillator-embedded automatic arrhythmia analysis algorithms. We propose a method for real-time ECG processing and parameter set extraction using band-pass digital filtration and ECG peak detection. Twelve parameters were derived: (i) seven parameters from the band-pass filter output-six threshold parameters and one peak counter; (ii) five parameters from the ECG peak detection branch, which assess the heart rate, the periodicity and the amplitude/slope symmetry of the ECG peaks. The statistical assessment for more than 36 h of cardiac arrhythmia episodes collected from the public AHA and MIT databases showed that some of the parameters achieved high specificity and sensitivity, but there was no parameter providing 100% separation between non-shockable and shockable rhythms. In order to estimate the influence of the wide variety of cardiac arrhythmias and the different artifacts in real recording conditions, we performed a more detailed study for eight non-shockable and four shockable rhythm categories. The combination of the six top-ranked parameters provided specificity: (i) more than 99% for rhythms with narrow supraventricular complexes, premature ventricular contractions, paced beats and bradycardias; (ii) almost 95% for supraventricular tachycardias; (iii) 91.5% for bundle branch blocks; (iv) 92.2% for slow ventricular tachycardias. The attained sensitivity was above 98% for coarse and fine ventricular fibrillations and 94% for the rapid ventricular tachycardias. The accuracy for the noise contaminated non-shockable and shockable signals exceeded 93%. The proposed parameter set guarantees an accuracy that meets the AHA performance goal for each rhythm category and could be a reliable facility for AED shock-advisory algorithms.

Cardiac arrhythmia classification using autoregressive modeling

BACKGROUND

Computer-assisted arrhythmia recognition is critical for the management of cardiac disorders. Various techniques have been utilized to classify arrhythmias. Generally, these techniques classify two or three arrhythmias or have significantly large processing times. A simpler autoregressive modeling (AR) technique is proposed to classify normal sinus rhythm (NSR) and various cardiac arrhythmias including atrial premature contraction (APC), premature ventricular contraction (PVC), superventricular tachycardia (SVT), ventricular tachycardia (VT) and ventricular fibrillation (VF).

METHODS

AR Modeling was performed on ECG data from normal sinus rhythm as well as various arrhythmias. The AR coefficients were computed using Burg's algorithm. The AR coefficients were classified using a generalized linear model (GLM) based algorithm in various stages.

RESULTS

AR modeling results showed that an order of four was sufficient for modeling the ECG signals. The accuracy of detecting NSR, APC, PVC, SVT, VT and VF were 93.2% to 100% using the GLM based classification algorithm.

CONCLUSION

The results show that AR modeling is useful for the classification of cardiac arrhythmias, with reasonably high accuracies. Further validation of the proposed technique will yield acceptable results for clinical implementation.

New Bayesian discriminator for detection of atrial tachyarrhythmias

BACKGROUND

Accurate, rapid detection of atrial tachyarrhythmias has important implications in the use of implantable devices for treatment of cardiac arrhythmia. Currently available detection algorithms for atrial tachyarrhythmias, which use the single-index method, have limited sensitivity and specificity.

METHODS AND RESULTS

In this study, we evaluated the performance of a new Bayesian discriminator algorithm in the detection of atrial fibrillation (AF), atrial flutter (AFL), and sinus rhythm (SR). Bipolar recording of 364 rhythms (AF=156, AFL=88, SR=120) at the high right atrium were collected from 20 patients who underwent electrophysiological procedures. After initial signal processing, a column vector of 5 features for each rhythm were established, based on the regularity, rate, energy distribution, percent time of quiet interval, and baseline reaching of the rectified autocorrelation coefficient functions. Rhythm identification was obtained by use of Bayes decision rule and assumption of Gaussian distribution. For the new Bayesian discriminator, the overall sensitivity for detection of SR, AF, and AFL was 97%, 97%, and 94%, respectively; and the overall specificity for detection of SR, AF, and AFL was 98%, 98%, and 99%, respectively. The overall accuracy of detection of SR, AF, and AFL was 98%, 97% and 98%, respectively. Furthermore, sensitivity, specificity, and accuracy of this algorithm were not affected by a range of white Gaussian noises with different intensities.

CONCLUSIONS

This new Bayesian discriminator algorithm, based on Bayes decision of multiple features of atrial electrograms, allows rapid on-line and accurate (98%) detection of AF with robust anti-noise performance.