Search

In this paper, we propose an end-to-end approach to addressing QRS complex detection and measurement of Electrocardiograph (ECG) paper using convolutional neural networks (CNNs). Unlike conventional detection solutions that convert images to digital data, our method can directly detect QRS complex in images using Faster-RCNN, then the R-peak can be located and measured through a CNN. Validated by clinical ECG data in the St.-Petersburg Institute of Cardiological Technics 12-lead Arrhythmia Database and real ECG paper from Peking University People's Hospital, the proposed method can achieve the recall of 98.32%, the precision of 99.01% in detecting and 0.012 mv of mean absolute error in measuring. Experimental results demonstrate the superior performance of our method over conventional solutions, which would pave the way to detect and measure ECG paper using CNNs.

The importance of fetal surveillance during pregnancy is worldwide accepted since its peculiar ability to anticipate fetal distress under a variety of conditions. The novel frontier in the field of remote fetal monitoring relies on a continuous and everyday-monitoring of fetal wellbeing. As a consequence, fECG monitoring systems have seen a net increase in popularity in the recent years. In this paper, we propose a novel algorithm for the detection of fECG and we validated its performances by testing it on an open source collection of 75 annotated fECG traces. Our results show the reliability of the proposed methodology in extracting fECG and deriving an estimate of fHR.

T-wave amplitude (TWA) is a well know index of the autonomic innervation of the myocardium. However, until now it has been evaluated only manually or with simple and inefficient algorithms. In this paper, we developed a new robust single-lead electrocardiogram (ECG) T-wave delineation algorithm that is able to detect the T-wave with a wavelet based method and automatically calculate the TWA. We evaluated the algorithm on the QT database, achieving a sensitivity of 99.92% for the T wave peak and 99.38% for the T wave end. In addition, the percentage of records automatically delineated with high precision was higher than previous published works. Finally, the algorithm was applied to study the influence of anticholinergic and antiadrenergic drugs (i.e. atropine and metoprolol) on the TWA. It was observed that atropine significantly decreased the TWA when compared to baseline level, that head-up tilt caused a decrease of TWA and that metoprolol blunted this decrease. Through the development of a robust algorithm, this study opens the way for further research on the T-wave analysis for the assessment of the autonomic innervation of the ventricular myocardium.

In this paper the application of statistical approach for anomaly detection in a wearable ECG monitor has been considered. The method is based on the usage of the pre-defined expected behavior of the monitored biomedical signal and its on-line comparison with real-time measurements. Such comparison can be implemented based on the statistical model of ECG signal. To test the accuracy of the proposed method a ProSim 8 simulator was used to generate ECG signals. Our experiments showed that critical anomalies in ECG, such as Cardiac Failure, different types of Arrhythmia, and ST-segment deviations can be detected with high precision (96~100%), while false positive rate over typical NSR signal is low (<;4%). The method is scalable in terms of required performance, power consumption, and the precision of anomaly detection to be suitable for the implementation in a wearable platform.

Electrocardiogram (ECG) is a widely used non-invasive method to study the rhythmic activity of the heart and thereby to detect the abnormalities. However, these signals are often obscured by artifacts from various sources and minimization of these artifacts are of paramount important. This paper proposes two adaptive techniques, namely the EEMD-BLMS (Ensemble Empirical Mode Decomposition in conjunction with the Block Least Mean Square algorithm) and DWT-NN (Discrete Wavelet Transform followed by Neural Network) methods in minimizing the artifacts from recorded ECG signals, and compares their performance. These methods were first compared on two types of simulated noise corrupted ECG signals: Type-I (desired ECG+noise frequencies outside the ECG frequency band) and Type-II (ECG+noise frequencies both inside and outside the ECG frequency band). Subsequently, they were tested on real ECG recordings. Results clearly show that both the methods works equally well when used on Type-I signals. However, on Type-II signals the DWT-NN performed better. In the case of real ECG data, though both methods performed similar, the DWT-NN method was a slightly better in terms of minimizing the high frequency artifacts.

In this paper we propose a novel approach for estimating narrowband components from bioelectrical signals. The approach is based on the notion of modulated quadratic variation, introduced as a measure of variability for narrowband signals. The algorithm is the closed-form solution to a constrained convex optimization problem, where narrowband components are estimated tracking the slow variations around a central frequency in the measured signal. The approach is general and can be applied to any bioelectrical signal, either for diagnostic or denoising purposes. In this paper we assess its performance on ECG and EMG signals. Numerical results show its effectiveness in removing narrowband artifacts, such as power-line interference, while preserving signal morphology. It greatly outperforms conventional notch filtering. Moreover, it is also very fast, as its computational complexity is linear in the size of the vector to process.

Sleep apnea is a sleep disorder distinguished by repetitive absence of breathing. Compared with the traditional expensive and cumbersome methods, sleep apnea diagnosis or screening with physiological information that can be easily acquired is needed. This paper describes algorithms using heart rate variability (HRV) to automatically detect sleep apneas as long as it can be easily acquired with unobtrusive sensors. Because the changes in cardiac activity are usually hysteretic than the presence of apneas with a few minutes, we propose to use the delayed HRV features to identify the episodes with sleep apneic events. This is expected to help improve the apnea detection performance. Experiments were conducted with a data set of 23 sleep apnea patients using support vector machine (SVM) classifiers and cross validations. Results show that using eleven HRV features with a time delay of 1.5 minutes rather than the features without time delay for SA detection, the overall accuracy increased from 74.9% to 76.2% and the Cohen's Kappa coefficient increased from 0.49 to 0.52. Further, an accuracy of 94.5% and a Kappa of 0.89 were achieved when applying subject-specific classifiers.

Recently it was shown that the spatial dispersion of ventricular repolarization (SHVR) can be assessed from the surface ECG using a metric termed ν-index. In this paper, a new algorithm is presented for estimating the ν-index, allowing the inclusion of higher order terms with ease, even in the presence of noise, leading to more accurate estimates. We first introduced a new analytical model for the derivative of the average transmembrane potentials during repolarization (the dominant T-wave) based on trigonometric functions. This functional set is closed under the operation of derivation. Therefore, the nonlinear iterative optimization required by previous methods is no longer necessary. Then, we suggested an iterative linear matrix factorization method to properly estimate the leads factors and the ν-index. Several synthetic SHVR (in the range 20 to 70 ms) were simulated, employing a publicly-available forward electrophysiological model (ECGSIM), leading to a total of 240 synthetic 8-lead electrocardiographical recordings (ECG), each composed of 128 beats. Then the ν-index was estimated with the newly introduced method and compared (root mean square error, RMSE) with the theoretical values, available for each series. The simulation results confirmed the theoretical expectations and indeed showed that the ν-index estimates were improved by increasing the number of lead factors included (RMSE=0:295±0:037 vs 0:280±0:038 for 2 and 8 lead factors respectively).

QRS complex detecting algorithm is core of ECG auto-diagnosis method and deeply influences cardiac cycle division for signal compression. However, ECG signals collected by noninvasive surface electrodes areusually mixed with several kinds of interference, and its waveform variation is the main reason for the hard realization of ECG processing. This paper proposes a QRS complex detecting algorithm based on multi-resolution mathematical morphological decomposition. This algorithm possesses superiorities in R peak detection of both mathematical morphological method and multi-resolution decomposition. Moreover, a lifting constructing method with Maximizationupdating operator is adopted to further improve the algorithm performance. And an efficient R peak search-back algorithm is employed to reduce the false positives (FP) and false negatives (FN). The proposed algorithm provides a good performance applying to MIT-BIH Arrhythmia Database, and achieves over 99% detection rate, sensitivity and positive predictivity, respectively, and calculation burden is low. Therefore, the proposed method is appropriate for portable medical devices in Telemedicine system.

This paper presents a computationally efficient QRS detection algorithm for wearable electrocardiogram (ECG) applications based on dual-slope analysis. In general, ECG signals of arrhythmias are pseudo-periodic and contaminated with noises like the patient's contraction muscles, respiration, 60 Hz interference and other types which impede correct QRS detection. To resolve this problem, in this paper, a technique is presented which is based on two slopes on both sides of a peak in ECG signal. Based on these slopes, a variable measuring steepness is developed and R peaks are detected. The algorithm was evaluated against MIT/BIH arrhythmia database and achieved 99.38% detection rate. This method was compared with one of the recently developed dual-slope based QRS detection methods. The results showed that the proposed method has 12.48 times faster runtime than the old method.

This paper addresses compressive sensing for multi-channel ECG. Compared to the traditional sparse signal recovery approach which decomposes the signal into the product of a dictionary and a sparse vector, the recently developed cosparse approach exploits sparsity of the product of an analysis matrix and the original signal. We apply the cosparse Greedy Analysis Pursuit (GAP) algorithm for compressive sensing of ECG signals. Moreover, to reduce processing time, classical signal-channel GAP is generalized to the multi-channel GAP algorithm, which simultaneously reconstructs multiple signals with similar support. Numerical experiments show that the proposed method outperforms the classical sparse multi-channel greedy algorithms in terms of accuracy and the single-channel cosparse approach in terms of processing speed.

Long-term recording of Electrocardiogram (ECG) signals plays an important role in health care systems for diagnostic and treatment purposes of heart diseases. Clustering and classification of collecting data are essential parts for detecting concealed information of P-QRS-T waves in the long-term ECG recording. Currently used algorithms do have their share of drawbacks: 1) clustering and classification cannot be done in real time; 2) they suffer from huge energy consumption and load of sampling. These drawbacks motivated us in developing novel optimized clustering algorithm which could easily scan large ECG datasets for establishing low power long-term ECG recording. In this paper, we present an advanced K-means clustering algorithm based on Compressed Sensing (CS) theory as a random sampling procedure. Then, two dimensionality reduction methods: Principal Component Analysis (PCA) and Linear Correlation Coefficient (LCC) followed by sorting the data using the K-Nearest Neighbours (K-NN) and Probabilistic Neural Network (PNN) classifiers are applied to the proposed algorithm. We show our algorithm based on PCA features in combination with K-NN classifier shows better performance than other methods. The proposed algorithm outperforms existing algorithms by increasing 11% classification accuracy. In addition, the proposed algorithm illustrates classification accuracy for K-NN and PNN classifiers, and a Receiver Operating Characteristics (ROC) area of 99.98%, 99.83%, and 99.75% respectively.

This paper proposes a new method for detecting P and T waves in multilead ECG based on the Finite Rate of Innovation(FRI) technique [8]. A simple QRS detection scheme will be presented followed by a novel P and T wave detection algorithm. The novelty here is the modelling of the P and T wave using a Gaussian kernel. Using a 2D wavelet decomposition, the approximation coefficients are windowed based on the QRS locations. The FRI method is then used to identify the Gaussian distribution present in the window which will in turn provide the locations of the P and T wave. This method was tested on more than an hour of clean and noisy data and shows good performance in the noisy case.

The electrocardiogram (ECG), widely known as a cardiac diagnostic signal, has recently been proposed for biometric identification of individuals; however reliability and reproducibility are of research interest. In this paper, we propose a template matching technique with 12 features using logistic regression classifier that achieved high reliability and identification accuracy. Non-invasive ECG signals were captured using our custom-built ambulatory EEG/ECG embedded device (NeuroMonitor). ECG data were collected from healthy subjects (10), between 25-35 years, for 10 seconds per trial. The number of trials from each subject was 10. From each trial, only 0.58 seconds of Lead I ECG data were used as template. Hardware-efficient fiducial point detection technique was implemented for feature extraction. To obtain repeated random sub-sampling validation, data were randomly separated into training and testing sets at a ratio of 80:20. Test data were used to find the classification accuracy. ECG template data with 12 extracted features provided the best performance in terms of accuracy (up to 100%) and processing complexity (computation time of 1.2ms). This work shows that a single limb (Lead I) ECG can robustly identify an individual quickly and reliably with minimal contact and data processing using the proposed algorithm.

The T and P waves of electrocardiogram signals are excellent indicators in the analysis and interpretation of cardiac arrhythmia. As such, the need to address and develop an accurate delineation technique for the detection of these waves is necessary. In this paper, we present a novel robust and adaptive T and P wave delineation method for real-time analysis and nonstandard ECG morphologies. The proposed method is based on ECG signal filtering, value estimation of different fiducial points, applying backward and forward search windows as well as adaptive thresholds. Simulations and evaluations prove the accuracy of the proposed technique in comparison to those proposed techniques in the literature. The mean error for the T peak, T offset, P peak and P offset values are found to be 9.8, 2.3, 7.3 and 3.5 milliseconds, respectively, based on the Physionet QT database, rendering our algorithm as an excellent candidate for ECG signal analysis.

Single Channel Blind Source Separation (SCBSS) is an extreme case of underdetermined (more sources and fewer sensors) Blind Source Separation (BSS) problem. In this paper, we propose a novel technique using Local Mean Decomposition (LMD) and Independent Component Analysis (ICA) combined with single channel BSS (LMD&#95;ICA). First, the LMD was used to decompose the single channel source into a series of data sequences, which are called as Product Functions (PF), then, ICA algorithm was used to process PFs to get similar independent components and extract the original signals. A comparison was made between LMD&#95;ICA and previously proposed single channel ICA method (EEMD&#95;ICA). The real time experimental results demonstrated the advantage of the proposed single channel source separation method for artifact removal and in biomedical source separation applications.

Observation of a patient's respiration signal can provide a clinician with the required information necessary to analyse a subject's wellbeing. Due to an increase in population number and the aging population demographic there is an increasing stress being placed on current healthcare systems. There is therefore a requirement for more of the rudimentary patient testing to be performed outside of the hospital environment. However due to the ambulatory nature of these recordings there is also a desire for a reduction in the number of sensors required to perform the required recording in order to be unobtrusive to the wearer, and also to use textile based systems for comfort. The extraction of a proxy for the respiration signal from a recorded electrocardiogram (ECG) signal has therefore received considerable interest from previous researchers. To allow for accurate measurements, currently employed methods rely on the availability of a clean artifact free ECG signal from which to extract the desired respiration signal. However, ambulatory recordings, made outside of the hospital-centric environment, are often corrupted with contaminating artifacts, the most degrading of which are due to subject motion. This paper presents the use of the ensemble empirical mode decomposition (EEMD) algorithm to aid in the extraction of the desired respiration signal. Two separate techniques are examined; 1) Extraction of the respiration signal directly from the noisy ECG 2) Removal of the artifact components relating to the subject movement allowing for the use of currently available respiration signal detection techniques. Results presented illustrate that the two proposed techniques provide significant improvements in the accuracy of the breaths per minute (BPM) metric when compared to the available true respiration signal. The error reduced from ± 5.9 BPM prior to the use of the two techniques to ± 2.9 and ± 3.3 BPM post processing using the EEMD algorithm techniques.

Constructing statistical models of electrocardiogram (ECG) signals, whose parameters can be used for automated disease classification, is of great importance in precluding manual annotation and providing prompt diagnosis of cardiac diseases. ECG signals consist of several segments with different morphologies (namely the P wave, QRS complex and the T wave) in a single heart beat, which can vary across individuals and diseases. Also, existing statistical ECG models exhibit a reliance upon obtaining a priori information from the ECG data by using preprocessing algorithms to initialize the filter parameters, or to define the user-specified model parameters. In this paper, we propose an ECG modeling technique using the sequential Markov chain Monte Carlo (SMCMC) filter that can perform simultaneous model selection, by adaptively choosing from different representations depending upon the nature of the data. Our results demonstrate the ability of the algorithm to track various types of ECG morphologies, including intermittently occurring ECG beats. In addition, we use the estimated model parameters as the feature set to classify between ECG signals with normal sinus rhythm and four different types of arrhythmia.

In this paper an efficient filtering procedure based on Extended Kalman Filter (EKF) has been proposed. The method is based on a modified nonlinear dynamic model, previously introduced for the generation of synthetic ECG signals. The proposed method considers the angular velocity of ECG signal, as one of the states of an EKF. We have considered two cases for observation equations, in one case we have assumed a corresponding observation to angular velocity state and in the other case, we have not assumed any observations for it. Quantitative evaluation of the proposed algorithm on the MIT-BIH Normal Sinus Rhythm Database (NSRDB) shows that an average SNR improvement of 8 dB is achieved for an input signal of -4 dB.

Discovering information encoded in non-invasively recorded biosignals which belies an individual's well-being can help facilitate the development of low-cost unobtrusive medical device technologies, or enable the unsupervised performance of physiological assessments without excessive oversight from trained clinical personnel. Although the unobtrusive or unsupervised nature of such technologies often results in less accurate measures than their invasive or supervised counterparts, this disadvantage is typically outweighed by the ability to monitor larger populations than ever before. The expected consequential benefit will be an improvement in healthcare provision and health outcomes for all. The process of discovering indicators of health in unsupervised or unobtrusive biosignal recordings, or automatically ensuring the validity and quality of such signals, is best realized when following a proven systematic methodology. This paper provides a brief tutorial review of supervised learning, which is a sub-discipline of machine learning, and discusses its application in the development of algorithms to interpret biosignals acquired in unsupervised or semi-supervised environments, with the aim of estimating well-being. Some specific examples in the disparate application areas of telehealth electrocardiogram recording and calculating post-operative systemic vascular resistance are discussed in the context of this systematic approach for information discovery.

Sleep apnea is the instance when one either has pauses of breathing in their sleep, or has very low breath while asleep. This pause in breathing can range in frequency and duration. Obstructive sleep apnea (OSA) is the common form of sleep apnea, which is currently tested through polysomnography (PSG) at sleep labs. PSG is both expensive and inconvenient as an expert human observer is required to work over night. New sleep apnea classification techniques are nowadays being developed by bioengineers for most comfortable and timely detection. This paper focuses on an automated classification algorithm which processes short duration epochs of the electrocardiogram (ECG) data. The presented classification technique is based on support vector machines (SVM) and has been trained and tested on sleep apnea recordings from subjects with and without OSA. The results show that our automated classification system can recognize epochs of sleep disorders with a high accuracy of 96.5% or higher. Furthermore, the proposed system can be used as a basis for future development of a tool for OSA screening.

Calculating entropy rate in physiologic signals has proven very useful in many settings. Common entropy estimates for this purpose are sample entropy (SampEn) and its less robust elder cousin, approximate entropy (ApEn). Both approaches count matches within a tolerance r for templates of length m consecutive observations. When physiologic data records are long and well-behaved, both approaches work very well for a wide range of m and r. However, more attention to the details of the estimation algorithm is needed for short records and signals with anomalies. In addition, interpretation of the magnitude of these estimates is highly dependent on how r is chosen and precludes comparison across studies with even slightly different methodologies. In this paper, we summarize recent novel approaches to improve the accuracy of entropy estimation. An important (but not necessarily new) alternative to current approaches is to develop estimates that convert probabilities to densities by normalizing by the matching region volume. This approach leads to a novel concept introduced here of reporting entropy rate in equivalent Gaussian white noise units. Another approach is to allow r to vary so that a pre-specified number of matches are found, called the minimum numerator count, to ensure confident probability estimation. The approaches are illustrated using a simple example of detecting abnormal cardiac rhythms in heart rate records.

The problem of detecting T-wave alternans (TWA) in ECG signals has received considerable attention in the biomedical community. This paper introduces a Bayesian model for the T waves contained in ECG signals. A block Gibbs sampler was recently studied to estimate the parameters of this Bayesian model (including wave locations, amplitudes and shapes). This paper shows that the samples generated by this Gibbs sampler can be used efficiently for TWA detection via different statistical tests constructed from odd and even T-wave amplitude samples. The proposed algorithm is evaluated on real ECG signals subjected to synthetic TWA and compared with two classical algorithms.

The aim of this paper was to study interactions between R-R intervals and respiratory phases in healthy children during night-time sleep using a novel technique based on joint symbolic dynamics. We investigated overnight polysomnography data of 40 healthy children. The R-R time series were extracted from electrocardiograms (ECG) and respiratory phases were obtained from abdominal sensors using the Hilbert transform. Both the series were transformed into ternary symbol vectors based on the changes between two successive R-R intervals or respiratory phases, respectively. Subsequently, words of length '2' were formed and the correspondence between words of the two series for each sleep stage was determined to quantify cardio-respiratory interaction. We found a significantly higher percentage of similarity in the joint symbolic dynamics of R-R intervals and respiratory phases during slow-wave (SW) sleep compared to any other sleep stage. There was, however, no significant effect of age, gender or BMI on cardio-respiratory interaction. In conclusion, joint symbolic dynamics provides a novel efficient technique for the analysis of cardio-respiratory interaction.

The Amplitude of Dominant T-Wave Alternans (ADTWA) is a recently introduced index which quantifies the presence of microvolt T Wave Alternans (TWA) on surface ECG recordings. In this paper we investigate the reliability of ADTWA and its robustness against broadband noise. At this regard, we generated synthetic 12-leads ECG recordings through a forward electrophysiological model and we added TWA, at different extent, by modulating the variation of the repolarization times of transmembrane action potentials across even and odd beats. Using a stochastic model, we derived an analytical relationship between the repolarization variation injected into the model and TWA at the surface, thus offering a strategy to evaluate lead sensitivity. In terms of robustness, the results of the simulations show that ADTWA correctly measured the amplitude of synthetic TWA with an average error of 3.3% ± 5.8% in absence of noise. When a 100 μV peak-to-peak broadband noise is present, its effects on estimation errors were kept limited by singular value decomposition on which ADTWA builds.

This paper investigates the viability of using the dynamic Bayesian Network framework as a tool to classify heart beats in long term ECG records. A Decision Support System composed by two layers is considered. The first layer performs the segmentation of each heartbeat available in the ECG record, whereas the second layer classifies the heartbeat as Premature Ventricular Contraction (PVC) or Other. The use of both static and dynamic Bayesian Networks is evaluated through using the records available in the MIT-BIH database, and the results show that the Dynamic one performs better, obtaining 95% of sensitivity and 98% of positive predictivity, showing that to consider the temporal relation among events is a good strategy to increase the certainty about present events.

An innovative electrocardiogram compression algorithm is presented in this paper. The proposed method is based on matrix completion, a new paradigm in signal processing that seeks to recover a low-rank matrix based on a small number of observations. The low-rank matrix is obtained via normalization of electrocardiogram records. Using matrix completion, the ECG data matrix is recovered from a few number of entries, thereby yielding high compression ratios comparable to those obtained by existing compression techniques. The proposed scheme offers a low-complexity encoder, good tolerance to quantization noise, and good quality reconstruction.

Electrocardiogram (ECG) is the P-QRS-T wave which indicates the electrical activity of the heart. The subtle changes in the amplitude and duration of the ECG signal depict the cardiac abnormality. It is very difficult to decipher these minute changes by the naked eye. Hence, a computer-aided diagnosis system will help the physicians to monitor the cardiac health. The ECG is a nonlinear and non-stationary signal. Hence, the hidden information in the ECG signal can be extracted using nonlinear method. In this paper, we have automatically classified normal and abnormal beats using higher order spectra (HOS) cumulants of wavelet packet decomposition (WPD). The abnormal beats are ventricular premature contractions (VPC) and Atrial premature contractions (APC). These HOS cumulant features of the WPD are subjected to principal component analysis (PCA) to reduce the number of features to five. Finally these features were fed to the support vector machine (SVM) with kernel functions for automatic classification. In our work, we have obtained the highest accuracy of 98.4% sensitivity and specificity of 98.9% and 98.0% respectively with radial basis function (RBF) kernel function and Meyer's wavelet (dmey) function. Our system is ready clinically to run on large amount of data sets.

Arrhythmia (i.e., irregular cardiac beat) classification in electrocardiogram (ECG) signals is an important issue for heart disease diagnosis due to the non-invasive nature of the ECG exam. In this paper, we analyze and criticize the results of some arrhythmia classification methods presented in the literature in terms of how the samples are chosen for training/testing the classifier and the impact this choice has on their performance (i.e., accuracy/sensitivity/specificity). From our implementation, we also report new accuracies for these methods, establishing a new state-of-the-art method, in terms of results.

Atrial fibrillation (A-fib) is the most common cardiac arrhythmia. To effectively treat or prevent A-fib, automatic A-fib detection based on Electrocardiograph (ECG) monitoring is highly desirable. This paper reviews recently developed techniques for A-fib detection based on non-episodic surface ECG monitoring data. A-fib detection methods in the literature can be mainly classified into three categories: (1) time domain methods; (2) frequency domain methods; and (3) non-linear methods. In general the performances of these methods were evaluated in terms of sensitivity, specificity and overall detection accuracy on the datasets from the Physionet repository. Based on our survey, we conclude that no promising A-fib detection method that performs consistently well across various scenarios has been proposed yet.

Analogue domain implementations of the Continuous Wavelet Transform (CWT) have proved popular in recent years as they can be implemented at very low power consumption levels. This is essential for use in wearable, long term physiological monitoring systems. Present analogue CWT implementations rely on taking mathematical a approximation of the wanted mother wavelet function to give a filter transfer function that is suitable for circuit implementation. This paper investigates the use of standard filter approximations (Butterworth, Chebyshev, Bessel) as an alternative wavelet approximation technique. This extends the number of approximation techniques available for generating analogue CWT filters. An example ECG analysis shows that signal information can be successfully extracted using these CWT approximations.

This paper examines the feasibility of accurate state classification of autonomic nervous activity (ANA) based on the power spectral pattern of the heart rate fluctuations (HRFs). Some attempts have been made to utilize artificial neural networks (ANNs) to classify HRFs for clinical diagnoses such as ischemic cardiomyopathy, arrhythmia or sleep apnea. To establish the firm bases for making such clinical diagnoses, it may be important to examine the classification accuracy for the data in physiologically well defined conditions by e.g. application of autonomic blocking agents. In this paper the three layered perceptron has been trained by the heart rate data in variety of ANS states yielded by the application of Atropine and Propranolol to 14 healthy male subjects. Six state (control, atropine and propranolol for each of the spine and upright posture) classification based on power spectrum showed average sensitivity of 67.2% and specificity 91.2%. Four state (control, atropine, propranolol and double block for either spine or upright posture) resulted in the average classification sensitivity of 75.7% and specificity 95.5%. The paper revealed that entropy bandwidth and indices originated from characteristic oscillations of blood pressure change improve the classification accuracy.

This paper presents an approach for an asynchronous BMI proposed as a switching part of a tremor suppression system developed for real-time continuous conditions. The main purpose of this BMI-switch is to anticipate the execution of self-initiated movements performed after relatively long periods of inactivity. The performance indicators used for the detector validation are specially suited for the continuous characteristic of the paradigm used and it is demonstrated that our ERD-based bayesian classifier solution is a reliable option, detecting a high rate of positive cases and generating very few false positives during long intervals of inactivity. The subjects analyzed for our detector validation were patients with neurological tremor caused by different pathologies in order to assure the adaptability of our system.

Electrocardiographic T wave peak-to-end interval (TpTe) is one parameter of T wave morphology, which contains indicators for hypoglycaemia. This paper shows the corrected TpTe (TpTe(c)) interval as one of the inputs contributing to detect hypoglycaemia. Support vector machine (SVM) and fuzzy support vector machine (FSVM) utilizing radial basis function (RBF) are used as the classification methods in this paper. By comparing with the classification systems using inputs of corrected QT interval (QT(c)) and heart rate only, the results indicate that the inclusion of TpTec in combination with QTc and heart rate performs better in the detection of hypoglycaemia in terms of sensitivity, specificity and accuracy.

This paper describes the performance of beat detection and heart rate variability (HRV) feature extraction on electrocardiogram signals which have been compressed and reconstructed with a lossy compression algorithm. The set partitioning in hierarchical trees (SPIHT) compression algorithm was used with sixteen compression ratios (CR) between 2 and 50 over the records of the MIT/BIH arrhythmia database. Sensitivities and specificities between 99% and 85% were computed for each CR utilised. The extracted HRV features were between 99% and 82% similar to the features extracted from the annotated records. A notable accuracy drop over all features extracted was noted beyond a CR of 30, with falls of 10% accuracy beyond this compression ratio.

In this paper we propose a novel channel identification scheme for solving permutation indeterminacy introduced by Independent Component Analysis (ICA) for artifacts removal from recorded three channel ECG signals within the remote health monitoring environment. The proposed scheme does not depend on the definition of any specific artifact which is the case with the existing approach and therefore leads to more robust and generic solution to this problem. The proposed scheme has been validated using nine practical case studies and its robustness has been proved by comparison with the existing approach. Simulation results show that the proposed scheme works successfully for all the nine cases whereas the existing approach fails to identify the correct channel for four cases.

In this paper an over-complete discrete wavelet transform (OCDWT) algorithm, obtained by blending two wavelet transform implementations that is redundant wavelet transform and the Mallat's multiresolution decomposition, has been proposed to retrieve the time-varying characteristics of HRV under two different postures, supine and standing. The OCDWT algorithm is critically sub-sampled to a given level of decomposition, below which it is then fully sampled. Five subjects were included to investigate posture-related HRV. The results showed that the high frequency fluctuations are larger in supine and get significantly reduced in standing in comparison to low frequency variations. Moreover, the very low frequency heart beat fluctuations during supine were greater than during standing. Further a comparative analysis has also been made between the Mallat's and OCDWT implementation in order to show the superiority of proposed algorithm.

ECG denoising has always been an important issue in medical engineering. The purposes of denoising are reducing noise level and improving signal to noise ratio (SNR) without distorting the signal. This paper proposes a method for removing white Gaussian noise from ECG signals. The concepts of singularity and local maxima of the wavelet transform modulus were used for analyzing singularity and reconstructing the ECG signal. Adaptive thresholding was used to remove white Gaussian noise modulus maximum of wavelet transform and then reconstruct the signal.

Atrial fibrillation (AFib) is the most common cardiac arrhythmia. Areas in atrial tissue with complex fractionated atrial electrograms (CFAEs) are among others responsible for the maintenance of AFib. Those areas are ideal target sites for ablation to eliminate AFib and restore sinus rhythm. As CFAEs are associated with high fibrillatory frequency, automated identification of CFAEs with spectral analysis helps developing objective strategies for AFib ablation. While the application of current techniques is restricted, this paper introduces a new approach to determine characteristic frequencies during AFib. By using Teager's energy operator we calculate the signal envelope and study its spectrum after Fast Fourier Transformation. Harmonic analysis of distinctive peaks in the power spectrum is carried out to assess characteristic frequencies of a CFAE. While the currently available methods only find one dominant frequency in the spectrum of the signal, our method is capable to find multiple characteristic frequencies, if present. Since it is believed that during AFib the atrium is activated by one or multiple wavelets, our method opens new opportunities for investigation of multiple wavelets propagation.

Abdominal recordings of fetal ECG (fECG) have lower signal-to-noise ratio (SNR) as compared with invasive procedures. In this paper we have combined two previously proposed methods, one for extracting fECG, called piCA and the other, a transformation based on Hilbert transform to enhance the R-peaks. The combination of these methods seems to work well in situations of noisy data and fetal repositioning. Also a comparison is done by using ICA in order to extract the fetal signals. Performance of both methods is studied separately. Results show that applying the transformation on the components extracted with the use of piCA (after maternal ECG cancellation), had a very good performance. Also, as piCA ranks the components according to their resemblance to maternal ECG, it could be used in automatic fetal ECG extraction algorithms.

This paper presents a method for synchronous recording and analyzing both the electrogastrographic signal (EGG) and the heart rate variability signal (HRV). The electrogastrographic examination can be considered as a noninvasive method for an investigation of a stomach slow wave propagation. The four channel signal are non-invasively captured by the appropriately placed electrodes on the surface of the stomach. The EGG and electrocardiographic (ECG) signals, recorded simultaneously by means of the same electrodes and an amplifier, are separated by the proper digital filtration. In our work the EGG and ECG analysis is limited to calculation the most frequently used parameters: the dominant frequency of the EGG (DF), and LF/HF ratio of the HRV power spectrum for the ECG. In this way it is possible to examine mutual interaction among EGG and HRV. This paper also depicts the preliminary results of a comparison of the some EGG and the HRV parameters e.g. dominant power distribution (DPD) and balance of low frequency and high frequency of HRV power spectrum.

Numerous methods have been reported for deriving respiratory information such as respiratory rate from the electrocardiogram (ECG). In this paper the authors present a real-time algorithm for estimation and removal of baseline wander (BW) noise and obtaining the ECG-derived respiration (EDR) signal for estimation of a patient's respiratory rate. This algorithm utilizes a real-time "T-P knot" baseline wander removal technique which is based on the repetitive backward subtraction of the estimated baseline from the ECG signal. The estimated baseline is interpolated from the ECG signal at midpoints between each detected R-wave. As each segment of the estimated baseline signal is subtracted from the ECG, a "flattened" ECG signal is produced for which the amplitude of each R-wave is analyzed. The respiration signal is estimated from the amplitude modulation of R-waves caused by breathing. Testing of the algorithm was conducted in a pseudo real-time environment using MATLAB(TM), and test results are presented for simultaneously recorded ECG and respiration recordings from the PhysioNet/PhysioBank Fantasia database. Test data from patients were chosen with particularly large baseline wander components to ensure the reliability of the algorithm under adverse ECG recording conditions. The algorithm yielded EDR signals with a respiration rate of 4.4 breaths/min. for Fantasia patient record f2y10 and 10.1 breaths/min. for Fantasia patient record f2y06. These were in good agreement with the simultaneously recorded respiration data provided in the Fantasia database thus confirming the efficacy of the algorithm.

Presented paper describes a system of biomedical signal classifiers with preliminary feature extraction stage based on matched wavelets analysis, where two structures of classifier using Neural Networks (NN) and Support Vector Machine (SVM) are applied. As a pilot study the rules extraction algorithm applied for two of mentioned machine learning approaches (NN & SVM) was used. This was made to extract and transform the representation of knowledge gathered in Black Box parameters during classifier learning phase to be better and natural understandable for human user/expert. Proposed system was tested on the set of ECG signals of 20 atrial fibrillation (AF) and 20 control group (CG) patients, divided into learning and verifying subsets, taken from MIT-BiH database. Obtained results showed, that the ability of generalization of created system, expressed as a measure of sensitivity and specificity increased, due to extracting and selectively choosing only the most representative features for analyzed AF detection problem. Classification results achieved by means of constructed matched wavelet, created for given AF detection features were better than indicators obtained for standard wavelet basic functions used in ECG time-frequency decomposition.

This study investigates the rate estimation problem encountered in rehabilitation exercise monitoring by using noninvasive portable sensors. The purpose of this paper has two main parts. The first part is to find suitable approaches for the rate detection of tri-axial accelerometer (TA) signals and ECG signals respectively. It is found that the integral type approaches (the average magnitude difference function (AMDF) and autocorrelation function (ACF)) are particularly suitable for TA signal pre-processing, while differential type approaches are very efficient for electrocardiographic (ECG) signal pre-processing. The second part is to develop a square wave matching method to detect the rate from the pre-processed signals. Experimental results indicate that the proposed methods can effectively detect pace rate from TA and heart rate from ECG and remove undesirable spikes.

This paper describes a new approach in features extraction using time-frequency distributions (TFDs) for detecting epileptic seizures to identify abnormalities in electroencephalogram (EEG). Particularly, the method extracts features using the Smoothed Pseudo Wigner-Ville distribution combined with the McAulay-Quatieri sinusoidal model and identifies abnormal neural discharges. We propose a new feature based on the length of the track that, combined with energy and frequency features, allows to isolate a continuous energy trace from another oscillations when an epileptic seizure is beginning. We evaluate our approach using data consisting of 16 different seizures from 6 epileptic patients. The results show that our extraction method is a suitable approach for automatic seizure detection, and opens the possibility of formulating new criteria to detect and analyze abnormal EEGs.

A Brain-Computer Interface (BCI) is a system that could enable patients like those with Amyotrophic Lateral Sclerosis to control some equipment and to communicate with other people, and has been anticipated to be achieved. One of the problems in BCI research is a trade-off between speed and accuracy, and it is practically important to adjust those two performance measures effectively. So far the authors have considered BCIs as communications between users and computers, and have proposed an error control method, Reliability-Based Automatic Repeat reQuest (RB-ARQ). It has been shown that, with Linear Discriminant Analysis (LDA) as a classifier, RB-ARQ is more effective than other error control methods. In this paper, Support Vector Machines (SVMs), one of the most popular classifiers, are applied to RB-ARQ. A quantitative comparison showed that there was no significant difference between LDA and SVM. Also, it was demonstrated that RB-ARQ improved the accuracy from the one acquired by the top ranked methods in the BCI competition to 100 percents, with less loss of the speed.

To facilitate research and development in Brain-Computer Interface (BCI) research, we have been developing a general-purpose BCI system, called BCI2000, over the past nine years. This system has enjoyed a growing adoption in BCI and related areas and has been the basis for some of the most impressive studies reported to date. This paper gives an update on the status of this project by describing the principles of the BCI2000 system, its benefits, and impact on the field to date.

T wave alternans (TWA) has been associated with ventricular arrhythmias. Hence, TWA detection can risk stratify patients with heart disease who may experience sudden death from ventricular arhythmias. However, accurate TWA detection is technically challenging due to the low microvolt TWA signal and the confounding effect of biological noise such as movement, myopotentials or respiration. In this paper, we propose nonnegative matrix factorization (NMF)-Adaptive spectral method to increase the robustness of TWA detection in ambulatory electrocardiograms (ECGs). The proposed method applies a non-linear time-frequency (TF) analysis and NMF to the aligned ST-T waveforms. This method separates the TWA signal from the other non-desired ECG signal components, and detects TWA with high accuracy. The performance of our proposed method is validated in a clinical study using ECGs which confirms a TWA detection of 92% compared to 47% using the conventional spectral method.

This paper introduces a new approach to process biomedical signals by surgically removing wave deflections in time domain. The method first determines the epochs of high frequency deflections, cuts out them from the signal, and then connects the two disconnected points. To determine the epoch of a deflection to be removed, four slope trace waves are used to isolate the deflection based on signal characteristics of amplitude, slope, duration, and distance from neighboring deflections. The method has been applied to simulated data and MIT-BIH arrhythmia database to show its practical efficacy in the case of baseline wandering removal. It is found that the method has the capability to identify and remove high frequency deflections appropriately, leaving low frequency deflection such as baseline drifting.

In this paper, we introduce a new automatic method for electrocardiogram (ECG) artifact elimination from the electroencephalogram (EEG) or the electrooculogram (EOG). It is based on a modification of the independent component analysis (ICA) algorithm which gives promising results while only using a single-channel EEG (or EOG) and the ECG. To check the effectiveness of our approach, we compared its correction rate with those obtained by ensemble average subtraction (EAS) and adaptive filtering (AF). For this purpose, we applied these algorithms to 10 excerpts of polysomnographic sleep recordings containing ECG artifacts and other typical artifacts (e.g. movement, sweat, respiration, etc.). Two hundred successive interference peaks were examined in each excerpt to compute correction rates. We found that our modified ICA was the most robust to various waveforms of cardiac interference and to the presence of others artifacts, with a correction rate of 91.0%, against 83.5% for EAS and 83.1% for AF.