David Cuesta-Frau

A microcontroller-based portable electrocardiograph recorder

We describe a low cost portable Holler design that can be implemented with off-the-shelf components. The recorder is battery powered and includes a graphical display and keyboard. The recorder is capable of acquiring up to 48 hours of continuous electrocardiogram data at a sample rate of up to 250 Hz.

Characterization of Sample Entropy in the Context of Biomedical Signal Analysis

Sample entropy (SampEn) has been proposed as a method to overcome limitations associated with approximate entropy (ApEn). The initial paper describing the SampEn metric included a characterization study comparing both ApEn and SampEn against theoretical results and concluded that SampEn is both more consistent and agrees more closely with theory for known random processes than ApEn. SampEn has been used in several studies to analyze the regularity of clinical and experimental time series. However, questions regarding how to interpret SampEn in certain clinical situations and its relationship to classical signal parameters remain unanswered. In this paper we report the results of a characterization study intended to provide additional insights regarding the interpretability of SampEn in the context of biomedical signal analysis.

Clustering of electrocardiograph signals in computer-aided Holter analysis

A number of methods for temporal alignment, feature extraction and clustering of electrocardiographic signals are proposed. The ultimate aim of the paper is to find a method to automatically reduce the quantity of beats to examine in a long-term electrocardiographic signal, known as Holter signal, without loss of valuable information for the diagnosis. These signals include thousands of beats and therefore visual inspection is difficult and cumbersome. All the elements involved in each stage will be described and a thorough experimental study will be presented. The electrocardiograph signals used in the experiments belong to the well-known MIT database, where many different waveforms can be found. Based on the results of the experiments, a complete process is proposed to obtain the significant beats present within a signal, with a reasonable computational cost. Hence, cardiologists will only have to examine a small but fully representative subset of beats, making this method a very useful tool for medical decision support systems.

Comparative study of approximate entropy and sample entropy robustness to spikes

OBJECTIVE There is an ongoing research effort devoted to characterize the signal regularity metrics approximate entropy (ApEn) and sample entropy (SampEn) in order to better interpret their results in the context of biomedical signal analysis. Along with this line, this paper addresses the influence of abnormal spikes (impulses) on ApEn and SampEn measurements. METHODS A set of test signals consisting of generic synthetic signals, simulated biomedical signals, and real RR records was created. These test signals were corrupted by randomly generated spikes. ApEn and SampEn were computed for all the signals under different spike probabilities and for 100 realizations. RESULTS The effect of the presence of spikes on ApEn and SampEn is different for test signals with narrowband line spectra and test signals that are better modeled as broadband random processes. In the first case, the presence of extrinsic spikes in the signal results in an ApEn and SampEn increase. In the second case, it results in an entropy decrease. For real RR records, the presence of spikes, often due to QRS detection errors, also results in an entropy decrease. CONCLUSIONS Our findings demonstrate that both ApEn and SampEn are very sensitive to the presence of spikes. Abnormal spikes should be removed, if possible, from signals before computing ApEn or SampEn. Otherwise, the results can lead to misunderstandings or misclassification of the signal regularity.

Automatic Sleep Stages Classification Using EEG Entropy Features and Unsupervised Pattern Analysis Techniques

Sleep is a growing area of research interest in medicine and neuroscience. Actually, one major concern is to find a correlation between several physiologic variables and sleep stages. There is a scientific agreement on the characteristics of the five stages of human sleep, based on EEG analysis. Nevertheless, manual stage classification is still the most widely used approach. This work proposes a new automatic sleep classification method based on unsupervised feature classification algorithms recently developed, and on EEG entropy measures. This scheme extracts entropy metrics from EEG records to obtain a feature vector. Then, these features are optimized in terms of relevance using the Q-α algorithm. Finally, the resulting set of features is entered into a clustering procedure to obtain a final segmentation of the sleep stages. The proposed method reached up to an average of 80% correctly classified stages for each patient separately while keeping the computational cost low.

Unsupervised classification of ventricular extrasystoles using bounded clustering algorithms and morphology matching

Ventricular extrasystoles (VE) are ectopic heartbeats involving irregularities in the heart rhythm. VEs arise in response to impulses generated in some part of the heart different from the sinoatrial node. These are caused by the premature discharge of a ventricular ectopic focus. VEs after myocardial infarction are associated with increased mortality. Screening of VEs is typically a manual and time consuming task that involves analysis of the heartbeat morphology, QRS duration, and variations of the RR intervals using long-term electrocardiograms. We describe a novel algorithm to perform automatic classification of VEs and report the results of our validation study. The proposed algorithm makes use of bounded clustering algorithms, morphology matching, and RR interval length to perform automatic VE classification without prior knowledge of the number of classes and heartbeat features. Additionally, the proposed algorithm does not need a training set.

Unsupervised feature relevance analysis applied to improve ECG heartbeat clustering

The computer-assisted analysis of biomedical records has become an essential tool in clinical settings. However, current devices provide a growing amount of data that often exceeds the processing capacity of normal computers. As this amount of information rises, new demands for more efficient data extracting methods appear. This paper addresses the task of data mining in physiological records using a feature selection scheme. An unsupervised method based on relevance analysis is described. This scheme uses a least-squares optimization of the input feature matrix in a single iteration. The output of the algorithm is a feature weighting vector. The performance of the method was assessed using a heartbeat clustering test on real ECG records. The quantitative cluster validity measures yielded a correctly classified heartbeat rate of 98.69% (specificity), 85.88% (sensitivity) and 95.04% (general clustering performance), which is even higher than the performance achieved by other similar ECG clustering studies. The number of features was reduced on average from 100 to 18, and the temporal cost was a 43% lower than in previous ECG clustering schemes.

Speech recognition methods applied to biomedical signals processing

The paper focuses on processing of long biological signals used during monitoring procedures like in the case of portable Holter device for arrythmia analysis (ECG), intracranial pressure monitoring (ICP) in intensive care unit or overnight electroencephalogram monitoring (EEG) for sleep apnea detection. Two methods taken from speech processing are proposed: dynamic time warping (DTW) and hidden Markov models (HMM). The unsupervised analysis of ECG and ICP beats is carried out using hierarchical clustering approach. In case of EEG, first the estimation of sleep stages is performed and next the different breathing events are detected by HMM by means of Viterbi inference. We show that for the first two problems DTW outperforms HMM while in the third case the HMM inference capability makes HMM suitable for sleep apnea diagnosis.

Automatic segmentation of long-term ECG signals corrupted with broadband noise based on sample entropy

Biomedical signals are nonstationary in nature, namely, their statistical properties are time-dependent. Such changes in the underlying statistical properties of the signal and the effects of external noise often affect the performance and applicability of automatic signal processing methods that require stationarity. A number of methods have been proposed to address the problem of finding stationary signal segments within larger nonstationary signals. In this framework, processing and analysis are applied to each resulting locally stationary segment separately. The method proposed in this paper addresses the problem of finding locally quasi-stationary signal segments. Particularly, our proposed algorithm is designed to solve the specific problem of segmenting semiperiodic biomedical signals corrupted with broadband noise according to the various degrees of external noise power. It is based on the sample entropy and the relative sensitivity of this signal regularity metric to changes in the underlying signal properties and broadband noise levels. The assessment of the method was carried out by means of experiments on ECG signals drawn from the MIT-BIH arrhythmia database. The results were measured in terms of false alarms based on the changepoint detection bias. In summary, the results achieved were a sensitivity of 97%, and an error of 16% for records corrupted with muscle artifacts.

Influence of QRS complex detection errors on entropy algorithms. Application to heart rate variability discrimination

Signal entropy measures such as approximate entropy (ApEn) and sample entropy (SampEn) are widely used in heart rate variability (HRV) analysis and biomedical research. In this article, we analyze the influence of QRS detection errors on HRV results based on signal entropy measures. Specifically, we study the influence that QRS detection errors have on the discrimination power of ApEn and SampEn using the cardiac arrhythmia suppression trial (CAST) database. The experiments assessed the discrimination capability of ApEn and SampEn under different levels of QRS detection errors. The results demonstrate that these measures are sensitive to the presence of ectopic peaks: from a successful classification rate of 100%, down to a 75% when spikes are present. The discriminating capability of the metrics degraded as the number of misdetections increased. For an error rate of 2% the segmentation failed in a 12.5% of the experiments, whereas for a 5% rate, it failed in a 25%.