Classification of ischemic and non-ischemic cardiac events in Holter recordings based on the continuous wavelet transform.

Holter recordings are widely used to detect cardiac events that occur transiently, such as ischemic events. Much effort has been made to detect early ischemia, thus preventing myocardial infarction. However, after detection, classification of ischemia has still not been fully solved. The main difficulty relies on the false positives produced because of non-ischemic events, such as changes in the heart rate, the intraventricular conduction or the cardiac electrical axis. In this work, the classification of ischemic and non-ischemic events from the long-term ST database has been improved, using novel spectral parameters based on the continuous wavelet transform (CWT) together with temporal parameters (such as ST level and slope, T wave width and peak, R wave peak, QRS complex width). This was achieved by using a nearest neighbour classifier of six neighbours. Results indicated a sensitivity and specificity of 84.1% and 92.9% between ischemic and non-ischemic events, respectively, resulting a 10% increase of the sensitivity found in the literature. Extracted features based on the CWT applied on the ECG in the frequency band 0.5-4 Hz provided a substantial improvement in classifying ischemic and non-ischemic events, when comparing with the same classifier using only temporal parameters. Graphical Abstract In this work it is improved the classification of ischemic and non-ischemic events. The main difficulty of ischemic detectors relies on the false positives produced because of non-ischemic events. After a preprocessing stage, temporal and spectral parameters are extracted from events of the Long Term ST Database. The novel parameters proposed in this work are extracted from the Continuous Wavelet Transform. A nearest Neighbor Classifier is used, obtaining a sensitivity and specificity of 84.1% and 92.9%, respectively.

Quantification of Ventricular Repolarization Variation for Sudden Cardiac Death Risk Stratification in Atrial Fibrillation.

OBJECTIVE: Atrial fibrillation (AF) rhythm gives rise to an irregular response in ventricular activity, preventing the use of standard ECG-derived risk markers based on ventricular repolarization heterogeneity under this particular condition. In this study, we proposed new indices to quantify repolarization variations in AF patients, assessing their stratification performance in a chronic heart failure population with AF. METHODS: We developed a method based on a selective bin averaging technique. Consecutive beats preceded by a similar RR interval were selected, from which the average variation within the ST-T complex for each RR range was computed. We proposed two sets of indices: 1) the 2-beat index of ventricular repolarization variation, ( IV2), computed from pairs of stable consecutive beats; and 2) the 3-beat indices of ventricular repolarization variation, computed in triplets of stable consecutive beats ( IV3). RESULTS: These indices showed a significant association with sudden cardiac death (SCD) outcome in the study population. In addition, risk assessment based on the combination of the proposed indices improved stratification performance compared to their individual potential. CONCLUSION: Patients with enhanced ventricular repolarization variation computed in terms of the proposed indices were successfully associated to a higher SCD incidence in our study population, evidencing their prognostic value. SIGNIFICANCE: using a simple ambulatory ECG recording, it is possible to stratify AF patients at risk of SCD, which may help cardiologists in adopting most effective therapeutic strategies, with a positive impact in both the patient and healthcare systems.

Automatic classification of long-term ambulatory ECG records according to type of ischemic heart disease.

BACKGROUND: Elevated transient ischemic ST segment episodes in the ambulatory electrocardiographic (AECG) records appear generally in patients with transmural ischemia (e. g. Prinzmetal's angina) while depressed ischemic episodes appear in patients with subendocardial ischemia (e. g. unstable or stable angina). Huge amount of AECG data necessitates automatic methods for analysis. We present an algorithm which determines type of transient ischemic episodes in the leads of records (elevations/depressions) and classifies AECG records according to type of ischemic heart disease (Prinzmetal's angina; coronary artery diseases excluding patients with Prinzmetal's angina; other heart diseases). METHODS: The algorithm was developed using 24-hour AECG records of the Long Term ST Database (LTST DB). The algorithm robustly generates ST segment level function in each AECG lead of the records, and tracks time varying non-ischemic ST segment changes such as slow drifts and axis shifts to construct the ST segment reference function. The ST segment reference function is then subtracted from the ST segment level function to obtain the ST segment deviation function. Using the third statistical moment of the histogram of the ST segment deviation function, the algorithm determines deflections of leads according to type of ischemic episodes present (elevations, depressions), and then classifies records according to type of ischemic heart disease. RESULTS: Using 74 records of the LTST DB (containing elevated or depressed ischemic episodes, mixed ischemic episodes, or no episodes), the algorithm correctly determined deflections of the majority of the leads of the records and correctly classified majority of the records with Prinzmetal's angina into the Prinzmetal's angina category (7 out of 8); majority of the records with other coronary artery diseases into the coronary artery diseases excluding patients with Prinzmetal's angina category (47 out of 55); and correctly classified one out of 11 records with other heart diseases into the other heart diseases category. CONCLUSIONS: The developed algorithm is suitable for processing long AECG data, efficient, and correctly classified the majority of records of the LTST DB according to type of transient ischemic heart disease.

A personalized classification system for Holter registers.

In this paper we present a personalized long-term electrocardiogram (ECG) classification framework, which can be applied to any Holter register recorded from an individual patient. Due to the massive amount of ECG beats in a Holter register, visual inspection is quite difficult and cumbersome, if not impossible. Therefore the proposed system helps professionals to quickly and accurately diagnose any latent heart disease by examining only the representative beats (the so called master key-beats) each of which is automatically extracted from a time frame of homogeneous (similar) beats. We tested the system on a benchmark database where beats of each Holter register have been manually labeled by cardiologists. The selection of the right master key-beats is the key factor for achieving a highly accurate classification and thus we used exhaustive K-means clustering in order to find out (near-) optimal number of key-beats as well as the master key-beats. The classification process produced results that were consistent with the manual labels with over 99% average accuracy, which basically shows the efficiency and the robustness of the proposed system over massive data (feature) collections in high dimensions.

Assessment of the classification capability of prediction and approximation methods for HRV analysis.

The goal of this paper is to examine the classification capabilities of various prediction and approximation methods and suggest which are most likely to be suitable for the clinical setting. Various prediction and approximation methods are applied in order to detect and extract those which provide the better differentiation between control and patient data, as well as members of different age groups. The prediction methods are local linear prediction, local exponential prediction, the delay times method, autoregressive prediction and neural networks. Approximation is computed with local linear approximation, least squares approximation, neural networks and the wavelet transform. These methods are chosen since each has a different physical basis and thus extracts and uses time series information in a different way.

Circadian rhythm of the QT interval dispersion in healthy subjects. Correlation with heart rate variability circadian pattern.

Circadian variation of QT interval dispersion (QTd) and heart rate variability spectral indices was evaluated in healthy persons in 24-hour 3-lead electrocardiogram. Mean values, SD, and SD/mean were evaluated for 24 hours, each hour separately and in night, day, and morning periods. Table Curve 2D and multiple regression were applied to find correlations between parameters. In 50% of subjects, a significant negative correlation was revealed between QTd and HF. Also, in 50% of persons, a significant positive correlation was found between QTd and low frequency/high frequency. After adjustment for periods, correlations were only observed during morning hours. With Table Curve 2D, 2 models of correlations between QTd and HF were found. Multiple regression analysis revealed relations between mean QTd and R-R as well as mean QTd and HF. It is possible that it is sympathovagal balance, as reflected in heart rate variability, and not the tone of both autonomic components that affects QTd variability.

Post infarction risk stratification using the 3-D angle between QRS complex and T-wave vectors.

Present experience with prospective identification of patients who might benefit from prophylactic antiarrhythmic intervention is restricted to risk stratification using left ventricular ejection fraction (LVEF). The precision of LVEF-based identification of high risk patients is neither highly sensitive nor highly specific. This study investigated risk stratification of 466 survivors of acute myocardial infarction (86 women, mean age 57.5 years) for whom a 5-year follow-up was available. During the follow-up 67 patients died and 24 of these events were sudden arrhythmic deaths. In addition to LVEF, patients were stratified by mean heart rate, heart rate variability and the slope of heart rate turbulence, all derived from 24-hour Holter recording obtained before hospital discharge, and by the 3D angle between QRS complex and T wave vectors (total cosine R-to-T, TCRT) obtained from digital resting electrocardiogram before hospital discharge. Individual risk characteristics and their combinations were evaluated by calculating the areas under the receiver operator characteristics (ROC). The bootstrap technology was used to investigate these statistically. For the stratification of both all cause mortality and sudden arrhythmic death, TCRT was the strongest risk stratifier (area under ROC of 0.6857 +/- 0.0367, and 0.7275 +/- 0.0544, respectively) that compared very favourably to LVEF (area under the ROC of 0.6610 +/- 0.0362 and 0.6346 +/- 0.0595, for all cause and arrhythmic death prediction, both P < 10(-10) for the comparison with TCRT). TCRT was also stronger in combination with other stratifiers, eg, TCRT + LVEF (area under ROC of 0.7631 +/- 0.0325 and 0.8057 +/- 0.0473, for all cause and arrhythmic death prediction) was stronger than mean heart rate + LVEF (area under ROC of 0.7396 +/- 0.0298 and 0.7673 +/- 0.0445, respectively, both P < 10(-10) for comparison with TCRT + LVEF). Hence the 3D QRS-T angle is a very powerful risk stratifier especially suited for the prediction of sudden arrhythmic death. It should be prospectively investigated in future trials of prophylactic antiarrhythmic interventions.

Characterization of QT interval adaptation to RR interval changes and its use as a risk-stratifier of arrhythmic mortality in amiodarone-treated survivors of acute myocardial infarction.

A new method is proposed to evaluate the dynamics of QT interval adaptation in response to heart rate (HR) changes. The method considers weighted averages of RR intervals (RR) preceding each cardiac beat to express RR interval history accounting for the influence on repolarization duration. A global optimization algorithm is used to determine the weight distribution leading to the lowest regression residual when curve fitting the [QT, RR1 data using a patient-specific regression model. From the optimum weight distribution, a memory lag L90 is estimated, expressing the delay in the QT adaptation to HR changes. On average, RR intervals of the past 150 beats (approximately 2.5 min) are required to model the QT response accurately. From a clinical point of view, the interval of the initial tens of seconds to one minute seems to be most important in the majority of cases. A measure of the optimum regression residual (r(opt)) has been calculated, discriminating between post-myocardial infarction patients at high and low risk of arrhythmic death while on treatment with amiodarone. A similar discrimination has been achieved with a variable expressing the character of QT lag behind the RR interval dynamics.

Dynamics of T-U wave in patients with idiopathic ventricular tachycardia originating from the right ventricular outflow tract.

Postextrasystolic U wave augmentation is observed in patients with long QT syndrome and those with organic heart disease. This phenomenon is considered a marker of increased risk of arrhythmia. However, the characteristics of the U wave have not been evaluated in patients with idiopathic VT originating from the right ventricular outflow tract (RVOT-VT). The present study evaluated the dynamic change in the T-U wave in patients with RVOT-VT. Holter ECGs obtained from 14 patients with RVOT-VT and 11 healthy control subjects were analyzed. The amplitude of T and U waves (Tamp and Uamp) and preceding RR intervals were measured during stable sinus rhythm (rate dependent change) and in the postextrasystolic sinus complex (pause dependent change). Uamp correlated negatively and significantly with the preceding RR interval in 13 (93%) RVOT-VT patients but in only 2 (18%) control subjects. The average value of the slope of the Uamp/RR relationship was negative (-0.22 +/- 0.10 mV/s) in the RVOT-VT group, but was positive (0.04 +/- 0.07 mV/s, P < 0.001) in the control group. Pause dependent U wave augmentation was observed in 12 (86%) of 14 patients. Increased frequency of consecutive preceding premature ventricular contractions (PVCs) was associated with a larger postextrasystolic Uamp. PVC or the first ventricular beat of VT arose from near the peak of augmented U waves. The dynamic changes in the T-U wave were observed in patients with RVOT-VT. Further investigations are required to elucidate the precise role of the U wave in arrhythmogenesis in those patients.

Simulation methods for the online extraction of ECG parameters under Matlab/Simulink.

The classification of cardiac pathologies in the human ECG greatly depends on the reliable extraction of characteristic features. This work presents a complete simulation environment for testing ECG classification algorithms under Matlab/Simulink. Evaluation of algorithm performance is undertaken in full compliance with the ANSI/AAMI standards EC38 and EC57, and ranges from beat-to-beat analysis to the comparison of episode markers (e.g., for VT/VF detection algorithms). For testing the quality of waveform boundary detection, our own testing methods have been implemented in compliance with existing literature.

Graphical representation of complex data--diurnal patterns of initiations of atrial fibrillation episodes.

A construction of a purpose designed graphical display is demonstrated in a study investigating the circadian distribution of patterns of RR interval sequences preceding episodes of paroxysmal atrial fibrillation (PAF). Based on a comparison with a (80%, 120%) range around the median of preceding 10 RR intervals, each RR interval is classified as normal, short, or long. Classifications of RR intervals in n-tuplets (n = 1, ...,5) preceding PAF episodes are used to compute probabilities of individual types of sequences occurring within 4-hour periods of the day (between 1 am, 5 am, 9 am, 1 pm, 5 pm, and 9 pm). Graphical representation of the data is proposed using a hierarchy of bar graphs. The graphical system has been filled with data of 327 atrial fibrillation episodes recorded in 46 24-hour ECGs in PAF patients. The graphical analysis supports a link between PAF initiation and cardiac autonomic status.

An intelligent system for the tracking--localization of changes & exploratory analysis--of long-term ECG.

An interactive methodology for the analysis of long-term ECG is introduced. It is an anthropomimetic technique and consists of three parts. At the first stage, the clinicians' scan of the ECG traces is imitated and changes in the shape of QRS are quantified. In the sequel, a clinician involves in the interpretation of the most prominent changes providing the patient-dependent prototypes for the subsequent machine learning procedure. Finally, a classification scheme incorporates the portion of medical knowledge needed to explore the whole patient's ECG. This scheme, being very robust to noise, presents excellent generalization properties and can serve as a reliable automation in a future examination of the certain subject.

Incidence of perioperative myocardial ischemia in TURP patients.

STUDY OBJECTIVE: To determine the incidence of new episodes of myocardial ischemia in patients undergoing transurethral resection of the prostate (TURP). DESIGN: Prospective, nonrandomized study. SETTING: Veterans Administration medical center. PATIENTS: 39 patients undergoing elective TURP. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Myocardial ischemia was detected with a 3-channel ambulatory ECG recorded. The ambulatory ECG recorder was applied preoperatively and removed when the patient left the recovery room. New myocardial ischemia was defined as a 1 mm or greater ST depression or a 2 mm or greater ST elevation from baseline, lasting for 1 minute or longer in at least one lead at the J point plus 60 msec unless this point fell within the T wave, in which case the J point 40 msec or greater was used. ST changes consistent with myocardial ischemia were confirmed by a cardiologist blinded to the patient's clinical course. Seven of 39 TURP patients (18%) had ST segment changes indicative of new myocardial ischemia. These seven patients had more prostate tissue resected and more blood loss than the 32 patients who did not have any myocardial ischemia (p < 0.05). CONCLUSIONS: Patients undergoing TURP have an 18% incidence of myocardial ischemia. Patients undergoing TURP with more prostate tissue resected and greater blood loss are at increased risk for perioperative myocardial ischemia.

Influence of recognition errors of computerised analysis of 24-hour electrocardiograms on the measurement of spectral components of heart rate variability.

Spectral methods for the assessment of heart rate variability (HRV) in 24-h electrocardiograms (ECG) are believed to require visual verification and manual editing of the computerised recognition of the ECG. This study investigated the effect of the recognition errors of computerised ECG recognition on two methods providing spectral HRV indices: (a) Fast Fourier Transformation (FFT); and (b) peak-to-trough analysis (PTA). Both methods were used to measure HRV spectra in 24-h ECGs recorded in 557 survivors of acute myocardial infarction. Each ECG was analysed using the Marquette 8000 Holter system and spectral HRV analyses were performed both prior to and after manual verification of the automatic ECG analysis. The FFT and PTA methods were used to calculate the low (0.04-0.15 Hz), medium (0.15-0.40 Hz) and high (0.40-1.00 Hz) HRV spectral components. For each method and for each spectral component, the rank correlations between the results obtained from unedited and edited ECG recognition were calculated. The correlations between the corresponding spectral components provided by the FFT and PTA methods applied to the edited recognitions were also calculated. Both methods were substantially affected by recognition errors. The FFT method was more sensitive to the misrecognition than the PTA method. The inter-method correlations were higher for the high and medium spectral components than for the low spectral component. The study suggests that spectral HRV analysis should be performed only on carefully verified and manually corrected recognitions of long-term electrocardiograms.