Project

Indentification of cardiac arrhythmias using unsupervised pattern recognition techniques

Goal: To explore the benefit of using non-supervised approaches for data representation and classification on the automatic identification of cardiac arrhythmias in ECG Holter recordings.

Updates
0 new
0
Recommendations
0 new
0
Followers
0 new
5
Reads
0 new
103

Project log

Diego Peluffo
added 5 research items
Cardiac arrhythmia analysis on Holter recordings is an important issue in clinical settings, however such issue implicitly involves attending other problems related to the large amount of unlabelled data which means a high computational cost. In this work an unsupervised methodology based in a segment framework is presented, which consists of dividing the raw data into a balanced number of segments in order to identify fiducial points, characterize and cluster the heartbeats in each segment separately. The resulting clusters are merged or split according to an assumed criterion of homogeneity. This framework compensates the high computational cost employed in Holter analysis, being possible its implementation for further real time applications. The performance of the method is measure over the records from the MIT/BIH arrhythmia database and achieves high values of sensibility and specificity, taking advantage of database labels, for a broad kind of heartbeats types recommended by the AAMI.
Unsupervised pattern recognition analysis is the most used approach to grouping heartbeats of electrocardiographic recordings or electrocardiograms (ECGs). This is due to the fact that beats labeling are very often not available. Given that detection of some transient and infrequent arrhythmias is unfeasible in a short-time ECG test, ambulatory electrocardiography is required. In this paper, a design of a complete system for the identification of arrhythmias using unsupervised pattern recognition techniques is proposed. Particularly, our system involves stages for signal preprocessing, heartbeat segmentation and characterization, features selection, and clustering. All these stages are developed within a segment clustering framework, which is a suitable alternative to detect minority classes. As average performance, including five types of arrhythmia, our system reaches 91,31% and 99,16% for sensitivity and specificity, respectively.
Cardiac arrhythmia analysis on Holter recordings is an important issue in clinical settings, however such issue implicitly involves attending other problems related to the large amount of unlabelled data which means a high computational cost. In this work an unsupervised methodology based in a segment framework is presented, which consists of dividing the raw data into a balanced number of segments in order to identify fiducial points, characterize and cluster the heartbeats in each segment separately. The resulting clusters are merged or split according to an assumed criterion of homogeneity. This framework compensates the high computational cost employed in Holter analysis, being possible its implementation for further real time applications. The performance of the method is measure over the records from the MIT/BIH arrhythmia database and achieves high values of sensibility and specificity, taking advantage of database labels, for a broad kind of heartbeats types recommended by the AAMI.
Diego Peluffo
added a research item
This paper proposes an approach for modeling cardiac pulses from electrocardiographic signals (ECG). A modified van der Pol oscillator model (mvP) is analyzed, which, under a proper configuration, is capable of describing action potentials, and, therefore, it can be adapted for modeling a normal cardiac pulse. Adequate parameters of the mvP system response are estimated using non-linear dynamics methods, like dynamic time warping (DTW). In order to represent an adaptive response for each individual heartbeat, a parameter tuning optimization method is applied which is based on a genetic algorithm that generates responses that morphologically resemble real ECG. This feature is particularly relevant since heartbeats have intrinsically strong variability in terms of both shape and length. Experiments are performed over real ECG from MIT-BIH arrhythmias database. The application of the optimization process shows that the mvP oscillator can be used properly to model the ideal cardiac rate pulse.
Diego Peluffo
added 2 research items
In this work, an efficient non-supervised algorithm for clustering of ECG signals is presented. The method is assessed over a set of records from MIT/BIH arrhythmia database with different types of heartbeats, including normal (N) heartbeats, as well as the arrhythmia heartbeats recommended by the AAMI, usually found in Holter recordings: ventricular extra systoles (VE), left and right branch bundles blocks (LBBB and RBBB) and atrial premature beats (APB). The results are assessed by means the sensitivity and specificity measures, taking advantage of the database labels. Also, unsupervised performance measures are used. Finally, the performance of the algorithm is in average 95%, improving results reported by previous works of the literature.
Sleep stage classification is a highly addressed issue in polysomnography; It is considered a tedious and time-consuming task if done manually by the specialist; therefore, from the engineering point of view, several methods have been proposed to perform an automatic sleep stage classification. In this paper an unsupervised approach to automatic sleep stage clustering of EEG signals is proposed which uses spectral features related to signal power, coherences, asymmetries, and Wavelet coefficients; the set of features is classified using a clustering algorithm that optimizes a cost function of minimum sum of squares. Accuracy and kappa coefficients are comparable to those of the current literature as well as individual stage classification results. Methods and results are discussed in the light of the current literature, as well as the utility of the groups of features to differentiate the states of sleep. Finally, clustering techniques are recommended for implementation in support systems for sleep stage scoring.
Diego Peluffo
added 2 research items
This work presents an approach for modelling cardiac pulse from electrocardiographic signals (ECG). We explore the use of the Bonhoeffer-van der Pol (BVP) model-being a generalized version of the van der Pol oscillator-which, under proper parameters, is able to describe action potentials, and it can be then adapted to modelling normal cardiac pulse. Using basics of non-linear dynamics and some algebra, the BVP system response is estimated. To account for an adaptive response for every single heartbeat, we propose a parameter tuning method based on a heuristic search in order to yield responses that morphologically resemble real ECG. This aspect is important since heartbeats have intrinsically strong variability in terms of both shape and length. Experiments are carried out over real ECG from MIT-BIH arrhythmias database. We perform a bifurcation and phase portrait analysis to explore the relationship between non-linear dynamics features and pathology. Preliminary results provided here are promising showing some hints about the ability of non-linear systems modelling ECG to characterize heartbeats and facilitate the classification thereof, being latter very important for diagnosing purposes.
This work presents a comparative study of different partitional and spectral clustering techniques to cluster heartbeats patterns of long-term ECG signals. Due to the nature of signals and since, in many cases, it is not feasible labeling thereof, clustering is preferred for analysis. The use of a generic model of partitional clustering and the appropriate estimation of initialization parameters via spectral techniques represent some of the most important contributions of this research. The experiments are done with a standard arrhythmia database of MIT (Massachusetts Institute of Technology) and the feature extraction is carried out using techniques recommended by literature. Another important contribution is the design of a sequential analysis method which reduces the computational cost and improves clustering performance compared to traditional analysis that is, analyzing the whole data set in one iteration. Additionally, it suggests a complete system for unsupervised analysis of ECG signals, including feature extraction, feature selection, initialization and clustering stages. Also, some appropriate performance measures based on groups analysis were designed, which relate the clustering performance with the number of resultants groups and computational cost. This study is done taking into account the AAMI standard (Association for the Advance of Medical Instrumentation).
Jose LUIS Rodriguez-Sotelo
added 3 research items
Clustering is advisable technique for analysis and interpretation of long-term ECG Holter records. As a non-supervised method, several challenges are posed due to factors such as signal length (very long duration), noise presence, dynamic behavior and morphology variability (different patient physiology and/or pathology). This work describes an improved version of the k-means clustering algorithm (J-means) for this task. In order to reduce the number of heartbeats to process, a preclustering stage is also employed. Dissimilarity measure calculation is based on the Dynamic Time Warping approach. To assess the validity of the proposed method, a comparative study is carried out, using k-means, k-medians, hk-means, and J-means. Heartbeat features are extracted by means of WT coefficients and trace segmentation. Best results were achieved by the J-means algorithm, which reduces the clustering error down to 4.5% while the critical error tends to the minimal value.
The follow-up of some cardiac diseases may be achieved by ECG-holter record analysis. A heartbeat clustering method can be used to reduce the usually high computational cost of such Holter analysis. This study describes a method aimed at cardiac arrhythmia recognition based on this approach, by means of unsupervised inspection of morphologically similar heartbeat groups. Singular Value Decomposition (SVD) is used as the feature selection method since the complexity increases exponentially with the number of features. A modification of the k-means algorithm was developed for centroid computation, taking into account heartbeat length changes. Experimental set consisted of ECG records from the MIT database. The method yielded a 99.9% clustering accuracy considering pathological versus normal heartbeats. Both clustering error and critical error percentage was 0.01%.
ECG heartbeat type detection and classification are regarded as important procedures since they can significantly help to provide an accurate automated diagnosis. This paper addresses the specific problem of detecting atrial premature beats, that had been demonstrated to be a marker for stroke risk or cardiac arrhythmias. The proposed methodology consists of a stage to estimate characteristics such as morphology of P wave and QRS complex as well as indices of prematurity and a non-supervised stage used by the algorithm J-means to separate heartbeat feature vectors into classes. Partition initialization is carried out by a Max-Min approach. Experimental data set is taken from MIT-BIH arrhythmia database. Results evidence the reliability of the method since achieved sensitivity and specificity are high, 92.9 and 99.6%, respectively, for an average output number of 12 discovered clusters that can be considered as appropriate value to separate heartbeat classes from recordings.
Mónica Moreno
added 2 research items
An arrhythmia is a pathology that consists on altering the heartbeat. Although, the 12-lead electrocardiogram allows evaluation of the electrical behavior from heart to determine certain pathologies, there are some arrhythmias that are difficult to detect with this type of electrocardiography. In this sense, it is necessary the use of the Holter monitor because it facilitates the records of the heart electrical activity for long periods of time, it is usually 24 up to 48 hours. Due to the extension of the records provided by the monitor, it is common to use computational systems to evaluate diagnostic and morphological features of the beats in order to determine if there is any type of abnormality. These computational systems can be based on supervised or unsupervised pattern recognition techniques, however considering that the first option requires a visual inspection about the large number of beats present in a Holter record, it is an arduous task, as well as it involves monetary costs. Consequently, throughout this paper we present the design of a complete system for the identification of arrhythmias in Holter records using unsupervised pattern recognition techniques. The proposed system involves stages of pre-processing of the signal, segmentation and characterization of beats, as well as feature selection and clustering. In this case, the technique k-means is used. These steps are applied within the framework of a segment-based methodology that improves the detection of minority classes. Additionally, initialization criteria are considered, which allow to enhance quality measures, especially sensitivity. As a result, it is determined that using k-means with the max-min initialization and a number of groups equal to 12, it is possible to obtain the best results, with values of: 99.36%, 91.31% and 99.16% for accuracy, sensitivity and specificity, respectively.
An arrhythmia is an alteration of heart condition, which occurs due to the change of heart rate, mainly attributed to electric conduction system injury. Some arrhythmias, given its infrequent and transitory nature, are difficult to be detected with a 12-leads electrocardiography (ECG) test, and therefore ambulatory or Holter electrocardiography should take place, which allows for evaluating the patient for long periods of time without interfering with the patients’ daily activities [1]. In order to classify the representative heartbeat of the Holter recordings, techniques from either supervised or unsupervised analysis can be used; being the latter the most recommended [2]. Since the prior knowledge or labelling of known beats is mostly unfeasible, unsupervised techniques result advisable for this classification problem. Despite the existence of techniques than have been very helpful, the design of a robust system to face problems such as signal noise, the large amount of heartbeats, the minority classes, and the morphological variability [3] is still an open issue. The design of a complete system for the identification of arrhythmias is proposed using unsupervised techniques of representation and clustering, the system design involves stages for heartbeat segmentation, characterization, representation, evaluation of the sensibility of the number of clusters, clustering and evaluation of performance [4]. To characterize the heartbeats, morphological and spectral features that generate separability between cardiac arrhythmias are used and next the segment cluster is realized. The tests are made over recordings from MIT/BIH’s arrhythmia database, which includes 48 records with the arrhythmias recommended by the AAMI (Association for the advanced medical of instrumentation) such as : Normal beats (N), premature atrial beats (A), Premature ventricular contractions (V), right bundle branch block (R) and left bundle branch block (L). The number of features was reduced from 117 to number that depend of the record, to provide fair comparison among methods, the quality of grouping is measured by some unsupervised clustering quality indicators as accuracy, sensitivity and specificity. As average performance, including five types of arrhythmia, our system reaches 99,36%, 91,31,74% and 99,16% for accuracy, sensitivity and specificity, respectively.
Diego Peluffo
added a project goal
To explore the benefit of using non-supervised approaches for data representation and classification on the automatic identification of cardiac arrhythmias in ECG Holter recordings.