Ventricular Fibrillation and Tachycardia detection from surface ECG using time-frequency representation images as input dataset for machine learning.

BACKGROUND AND OBJECTIVE: To safely select the proper therapy for Ventricullar Fibrillation (VF) is essential to distinct it correctly from Ventricular Tachycardia (VT) and other rhythms. Provided that the required therapy would not be the same, an erroneous detection might lead to serious injuries to the patient or even cause Ventricular Fibrillation (VF). The main novelty of this paper is the use of time-frequency (t-f) representation images as the direct input to the classifier. We hypothesize that this method allow to improve classification results as it allows to eliminate the typical feature selection and extraction stage, and its corresponding loss of information. METHODS: The standard AHA and MIT-BIH databases were used for evaluation and comparison with other authors. Previous to t-f Pseudo Wigner-Ville (PWV) calculation, only a basic preprocessing for denoising and signal alignment is necessary. In order to check the validity of the method independently of the classifier, four different classifiers are used: Logistic Regression with L2 Regularization (L2 RLR), Adaptive Neural Network Classifier (ANNC), Support Vector Machine (SSVM), and Bagging classifier (BAGG). RESULTS: The main classification results for VF detection (including flutter episodes) are 95.56% sensitivity and 98.8% specificity, 88.80% sensitivity and 99.5% specificity for ventricular tachycardia (VT), 98.98% sensitivity and 97.7% specificity for normal sinus, and 96.87% sensitivity and 99.55% specificity for other rhythms. CONCLUSION: Results shows that using t-f data representations to feed classifiers provide superior performance values than the feature selection strategies used in previous works. It opens the door to be used in any other detection applications.

Sistema hardware de diagnostico de fibrilación ventricular en tiempo real mediante métodos de analisis tiempo-frecuencia / System hardware diagnostic of ventricular fibrillation in real time using methods of time-frequency analysis

Nowadays our society has to face different diseases related to cardiac pathologies which become more and more common. This is due to the daily life habits increasing the risk of suffering cardiac problems such as heart attack which could lead to death...

Application of adaptive signal processing for determining the limits of P and T waves in an ECG.

A new algorithm for the determination of the limits of P and T waves is proposed, and its foundations are mathematically analyzed. The algorithm performs an adaptive filtering so that the searched point corresponds to a minimum. Crucial properties of its performance are discussed, i.e., immunity to base line drifts and full adaptation to any cardiological criteria. A series of tests are made involving real registers with different morphologies for P and T-waves.

[The automatic, beat-to-beat determination of the QT interval and analysis of its spontaneous variability under basal conditions]. / Determinación automática, latido a latido, del intervalo QT y análisis de su variabilidad espontánea en condiciones basales.

OBJECTIVES: An analysis is made of the automatic beat-by-beat measurement of QT and other intervals related to ventricular repolarization. The variability pattern of these intervals is investigated in normal subjects at rest, along with their relation to RR cycle variability. MATERIAL AND METHODS: The electrocardiographic signals (LII) from 11 normal subjects (mean age 31 +/- 10 years) were recorded over 5 min and processed by applying specific algorithms to determine beat-by-beat the RR, QT, RT, QTm and RTm intervals (Tm = peak of T wave). An analysis was made of the variability of these intervals in the time (standard deviation, variation coefficient, difference between maximum and minimum values) and frequency domains (spectral analysis applying the Fourier transform). RESULTS: The differences between the automatic measurements and those performed by two observers (n = 110) were respectively -1.3 +/- 6.4 and -3.7 +/- 6.5 ms for QT, - 1.0 +/- 1.4 and -1.0 +/- 2.3 ms for QTm, -0.3 +/- 1.4 and -0.2 +/- 1.8 ms for RTm, and 0.7 +/- 6.5 and -2.8 +/- 10.3 ms for RT. The QT and RT intervals exhibited greater variability (SD = 6 +/- 1 ms) than QTm and RTm (SD = 3 +/- 1 ms, p < 0.0001). These differences persisted on comparing the corresponding variation coefficients. The differences between the maximum and minimum measurements were 45 +/- 24 ms for QT and RT, the values being significantly less in the case of QTm (21 +/- 26 ms, p < 0.05) and RTm (20 +/- 27 ms, p < 0.05). In the frequency domain, the high- (HF) and low-frequency (LF) band energies were low in the series formed by the ventricular repolarization intervals, and the LF band normalized amplitude was significantly lower than in the RR series. There were no significant differences in the frequencies of the maximum values of the LF and HF bands of the RR series with respect to the QT series. The correlations between the RR intervals and the subsequent repolarization intervals obtained in each subject were not significant in 7 of the 11 subjects studied. CONCLUSIONS: The automatic beat-by-beat determination of the ventricular repolarization intervals is precise, particularly when considering the intervals defined by the T wave peak. Repolarization variability during the sinus rhythm at rest is small, and is not linearly related to modifications of the previous RR interval. Neurovegetative and humoral influences are postulated to explain QT variations. The neurovegetative and humoral influences that regulate cardiac cycle and ventricular repolarization variability at rest, are found to be quantitatively different.