Felipe Alonso-Atienza

Detection of Life-Threatening Arrhythmias Using Feature Selection and Support Vector Machines

Early detection of ventricular fibrillation (VF) and rapid ventricular tachycardia (VT) is crucial for the success of the defibrillation therapy. A wide variety of detection algorithms have been proposed based on temporal, spectral, or complexity parameters extracted from the ECG. However, these algorithms are mostly constructed by considering each parameter individually. In this study, we present a novel life-threatening arrhythmias detection algorithm that combines a number of previously proposed ECG parameters by using support vector machines classifiers. A total of 13 parameters were computed accounting for temporal (morphological), spectral, and complexity features of the ECG signal. A filter-type feature selection (FS) procedure was proposed to analyze the relevance of the computed parameters and how they affect the detection performance. The proposed methodology was evaluated in two different binary detection scenarios: shockable (FV plus VT) versus nonshockable arrhythmias, and VF versus nonVF rhythms, using the information contained in the medical imaging technology database, the Creighton University ventricular tachycardia database, and the ventricular arrhythmia database. sensitivity (SE) and specificity (SP) analysis on the out of sample test data showed values of SE=95%, SP=99%, and SE=92% , SP=97% in the case of shockable and VF scenarios, respectively. Our algorithm was benchmarked against individual detection schemes, significantly improving their performance. Our results demonstrate that the combination of ECG parameters using statistical learning algorithms improves the efficiency for the detection of life-threatening arrhythmias.

Feature selection using support vector machines and bootstrap methods for ventricular fibrillation detection

Early detection of ventricular fibrillation (VF) is crucial for the success of the defibrillation therapy in automatic devices. A high number of detectors have been proposed based on temporal, spectral, and time-frequency parameters extracted from the surface electrocardiogram (ECG), showing always a limited performance. The combination ECG parameters on different domain (time, frequency, and time-frequency) using machine learning algorithms has been used to improve detection efficiency. However, the potential utilization of a wide number of parameters benefiting machine learning schemes has raised the need of efficient feature selection (FS) procedures. In this study, we propose a novel FS algorithm based on support vector machines (SVM) classifiers and bootstrap resampling (BR) techniques. We define a backward FS procedure that relies on evaluating changes in SVM performance when removing features from the input space. This evaluation is achieved according to a nonparametric statistic based on BR. After simulation studies, we benchmark the performance of our FS algorithm in AHA and MIT-BIH ECG databases. Our results show that the proposed FS algorithm outperforms the recursive feature elimination method in synthetic examples, and that the VF detector performance improves with the reduced feature set.

Nonuniform Interpolation of Noisy Signals Using Support Vector Machines

The problem of signal interpolation has been intensively studied in the information theory literature, in conditions such as unlimited band, nonuniform sampling, and presence of noise. During the last decade, support vector machines (SVM) have been widely used for approximation problems, including function and signal interpolation. However, the signal structure has not always been taken into account in SVM interpolation. We propose the statement of two novel SVM algorithms for signal interpolation, specifically, the primal and the dual signal model based algorithms. Shift-invariant Mercers kernels are used as building blocks, according to the requirement of bandlimited signal. The sine kernel, which has received little attention in the SVM literature, is used for bandlimited reconstruction. Well-known properties of general SVM algorithms (sparseness of the solution, robustness, and regularization) are explored with simulation examples, yielding improved results with respect to standard algorithms, and revealing good characteristics in nonuniform interpolation of noisy signals.

Fundamental Frequency and Regularity of Cardiac Electrograms With Fourier Organization Analysis

Dominant frequency analysis (DFA) and organization analysis (OA) of cardiac electrograms (EGMs) aims to establish clinical targets for cardiac arrhythmia ablation. However, these previous spectral descriptions of the EGM have often discarded relevant information in the spectrum, such as the harmonic structure or the spectral envelope. We propose a fully automated algorithm for estimating the spectral features in EGM recordings This approach, called Fourier OA (FOA), accounts jointly for the organization and periodicity in the EGM, in terms of the fundamental frequency instead of dominant frequency. In order to compare the performance of FOA and DFA-OA approaches, we analyzed simulated EGM, obtained in a computer model, as well as two databases of implantable defibrillator-stored EGM. FOA parameters improved the organization measurements with respect to OA, and averaged cycle length and regularity indexes were more accurate when related to the fundamental (instead of dominant) frequency, as estimated by the algorithm (p <; 0.05 comparing f0 estimated by DFA and by FOA). FOA yields a more detailed and robust spectral description of EGM compared to DFA and OA parameters.

Sensitivity and Spatial Resolution of Transvenous Leads in Implantable Cardioverter Defibrillator

It has been previously documented that the main features and sensing performance of electrograms (EGMs) recorded in implantable cardioverter defibrillators (ICDs) depend on lead configuration. Although this dependence has been ascribed to differences in lead sensitivity and spatial resolution, the quantification of these two properties on ICD has not yet been attempted. In this paper, an operative framework to study the spatial resolution of ICD transvenous leads is presented. We propose to quantify the spatial resolution of ICD transvenous leads based on a new characterization called lead resolution volume (ResV). We analyzed the sensitivity distribution and the ResV of two unipolar (tip-can and coil-can ) and two bipolar (true or tip-ring and integrated or tip-coil) ICD transvenous lead configurations. A detailed 3-D model of the human thorax based on the visible human man dataset was used to compute the lead sensitivity and computer simulations of simple cardiac dynamics were used to quantify the ResV. Differences in the sensitivity distribution throughout the ventricular myocardium (VM) were observed for each lead configuration. In our computer model of the human thorax, the ResV was found to comprise 7%, 35%, 45%, and 70% of the VM for true bipolar, integrated bipolar, tip-can unipolar, and coil-can unipolar ICD leads, respectively. Furthermore, our analysis shows that the spatial resolution depends on both lead sensitivity and cardiac dynamics, and therefore, it can vary for different heart rhythms.

Shape reconstruction of cardiac ischemia from non-contact intracardiac recordings: A model study

Abstract We present a new approach for the reconstruction of ischemic regions from only a few non-contact intracardiac recordings. Hence, it is desirable to exploit the spatio-temporal correlations contained in the data. To this end, we incorporate a time-dependent monodomain model of the cardiac electric activity into the inversion scheme. In order to take into account the electrophysiological alterations of ischemic regions, we also introduce appropriate variations of the model parameters. This approach allows us to perform the reconstruction of the affected regions successfully using only a few recording sites. The reconstruction process is based on level set techniques. Our numerical experiments in a bi-dimensional model of cardiac tissue validate our approach.

A Reliable Method for Rhythm Analysis during Cardiopulmonary Resuscitation

Interruptions in cardiopulmonary resuscitation (CPR) compromise defibrillation success. However, CPR must be interrupted to analyze the rhythm because although current methods for rhythm analysis during CPR have high sensitivity for shockable rhythms, the specificity for nonshockable rhythms is still too low. This paper introduces a new approach to rhythm analysis during CPR that combines two strategies: a state-of-the-art CPR artifact suppression filter and a shock advice algorithm (SAA) designed to optimally classify the filtered signal. Emphasis is on designing an algorithm with high specificity. The SAA includes a detector for low electrical activity rhythms to increase the specificity, and a shock/no-shock decision algorithm based on a support vector machine classifier using slope and frequency features. For this study, 1185 shockable and 6482 nonshockable 9-s segments corrupted by CPR artifacts were obtained from 247 patients suffering out-of-hospital cardiac arrest. The segments were split into a training and a test set. For the test set, the sensitivity and specificity for rhythm analysis during CPR were 91.0% and 96.6%, respectively. This new approach shows an important increase in specificity without compromising the sensitivity when compared to previous studies.

A student-centered collaborative learning environment for developing communication skills in engineering education

Communication skills development is one of the main goals of engineering education. We propose an integrated student-centered collaborative learning environment for developing communication skills, using project-based learning methods and peer assessment. In our learning environment, projects are assigned to small groups of students under teacher supervision, documented in a wiki-editing tool and presented during a public poster session. By combining wiki entries and poster presentations, we intend to facilitate students (1) to gain access to the project of their peers and share their results, (2) to analyze and comment critically the project of their peers and provide them with feedback, and (3) to enhance their writing and oral skills. Previous experiences encourage us to promote this integrated learning environment. Wiki environments allowed students to improve the quality of their projects and to develop a critical attitude towards their projects and the projects of their peers. The poster session was found to be more dynamic than traditional oral presentations. Students got engaged in a more open and critical manner with the project of their peers, and students presenting their project had the chance to improve the quality of their presentation on the fly, by presenting their work several times in the duration of the session. In future courses, we will implement a learning environment that combines both wiki-based and poster session approaches. We expect that the implementation of both approaches will help to develop the communication skills of engineering students.

Spectral analysis of intracardiac electrograms during induced and spontaneous ventricular fibrillation in humans

AIMS Very limited data are available on the differences between spontaneous and induced episodes of ventricular fibrillation (VF) in humans. The aim of the study was to compare the spectral characteristics of the electrical signal recorded by an implantable cardioverter defibrillator (ICD) during both types of episodes. METHODS AND RESULTS Thirteen ICD patients with at least one spontaneous and one induced VF recorded by the device were included in the study. A spectral representation was obtained for the first 3 s of the intracardiac unipolar electrogram during VF. The dominant frequency (f(d)), the peak power at f(d), an organization index (OI), a bandwidth measurement, and an estimate of the correlation with a sinusoidal wave (leakage) were estimated for each episode. The f(d) was higher in induced episodes (4.75 +/- 0.57 vs. 3.95 +/- 0.59 Hz for the spontaneous episodes, P = 0.002), as well as the degree of organization assessed by the OI, bandwidth, and leakage parameters. CONCLUSION Clinical and induced VF episodes in humans have different spectral characteristics. Changes in the electrophysiological substrate or in the location of the arrhythmia wavefront at onset could play a role to explain the observed differences.

Quality estimation of the electrocardiogram using cross-correlation among leads

BackgroundFast and accurate quality estimation of the electrocardiogram (ECG) signal is a relevant research topic that has attracted considerable interest in the scientific community, particularly due to its impact on tele-medicine monitoring systems, where the ECG is collected by untrained technicians. In recent years, a number of studies have addressed this topic, showing poor performance in discriminating between clinically acceptable and unacceptable ECG records.Methods This paper presents a novel, simple and accurate algorithm to estimate the quality of the 12-lead ECG by exploiting the structure of the cross-covariance matrix among different leads. Ideally, ECG signals from different leads should be highly correlated since they capture the same electrical activation process of the heart. However, in the presence of noise or artifacts the covariance among these signals will be affected. Eigenvalues of the ECG signals covariance matrix are fed into three different supervised binary classifiers.Results and conclusion The performance of these classifiers were evaluated using PhysioNet/CinC Challenge 2011 data. Our best quality classifier achieved an accuracy of 0.898 in the test set, while having a complexity well below the results of contestants who participated in the Challenge, thus making it suitable for implementation in current cellular devices.