The quantification of the QT-RR interaction in ECG signal using the detrended fluctuationanalysis and ARARX modelling.

In this paper, the detrended fluctuation analysis DFA is used to investigate and quantify the QT-RR interaction in different pathologic cases in order to distinguish between them. The study is carried out on the ECG signals of MIT-BIH universal database. Different ECG signals related to cardiac pathological cases are concerned with this study. These are: Premature Ventricular Contraction (PVC) (9 cases), Right Bundle Branch Block (RBBB) (4 cases), Left Bundle Branch Block (LBBB) (2 cases), Atrial Premature Beat (APB) (4 cases), Paced Beat (PB) (4 cases), and other pathologic cases with different severity (10 cases). All this cases are compared to the 15 normal cases. The obtained results show that the DFA can identify the healthy subject from the pathologic cases according to the values of the scaling exponent α. The results indicate that α varies between 0.5 and 1 in all cases which means that there is a long range correlation in RR and QT series. The QT and RR series are also modelled using the ARARX model. The parameters of the model are then extracted. The power spectral density (PSD) is estimated by using these parameters in order to provide further information about the causal interactions within the signals and also to determine the power scaling exponent ß. This scaling exponent confirms the relationship between RR and QT intervals in all the studied cases except in APB and PB cases where the behaviour is similar to that of the white noise. The QT variability degrees are calculated and the DFA is applied on it. The obtained results show a long range correlation between RR and QT intervals in all cases and an ambiguity in the APB case. The DFA is compared to the Poincaré method in order to evaluate the algorithm performance using the Fuzzy Sugeno classifier is used for this purpose.

Analysis of the QT-RR variability interactions using the NARMAX model.

In this paper a new approach is used in order to evaluate and quantify the interactions between the QT and RR intervals. This is achieved after the identification of the RR and QT series with a hybrid model (the non-linear autoregressive moving average with exogenous input (NARMAX)). This identification follows two steps: the first is a linear parametric identification corresponding to the MA model, whereas the second is a non-linear identification using the NARX model. The power spectral density PSD of RR and QT is computed by using the monovariate part of this model (MA model). The QT-related RR series is obtained by using the bivariate part corresponding to the NARX model and its PSD is determined by using the autoregressive method. Then a cross-spectral and the coherence function were determined in order to confirm the obtained results. Different heart pathology cases were selected to evaluate the approach: the normal case, the cases which represent long QT intervals and some other cases which represent short QT intervals. They were taken from the MIT BIH database. The results show that every case illustrates two frequencies; the first in the low frequency band LF and the second in the high frequency band HF. In the normal case and long QT interval cases, the LF was predominating in the QT, RR and in QT-related RR power spectral density PSD. In the short QT interval cases the HF was much larger in all cases. The obtained results were compared to the poincaré plot method which confirms it; however, the NARMAX model can distinguish between normal and pathologic cases with a great precision (p < 0.001). In addition, the QT variability index QTVI is computed and represented by a box plot which expresses the relationship between QT and RR intervals. The QTVI shows a large variability in the short QT interval cases, whereas it shows a small and a negative variability in the long QT interval case.

Detection of the QRS complex by linear prediction.

The electrocardiogram (ECG) represents the electrical activity of the heart. It is characterized by its recurrent or periodic behaviour with each beat. Each recurrence is composed of a wave sequence consisting of P, QRS and T-waves, where the most characteristic wave set is the QRS complex. In this paper, we have developed an algorithm for detection of the QRS complex. The algorithm consists of several steps: signal-to-noise enhancement, linear prediction for ECG signal analysis, nonlinear transform, moving window integrator, centre-clipping transformation and QRS detection. Linear prediction determines the coefficients of a forward linear predictor by minimizing the prediction error by a least-square approach. The residual error signal obtained after processing by the linear prediction algorithm has very significant properties which will be used to localize and detect QRS complexes. The detection algorithm is tested on ECG signals from the universal MIT-BIH arrhythmia database and compared with the Pan and Tompkins QRS detection method. The results we obtain show that our method performs better than this method. Our algorithm results in fewer false positives and fewer false negatives.

A Powerful Algorithm to aid Cardiac Arrhythmia Diagnosis.

The Electrocardiogram (ECG), represents the electrical activity of the heart. It is characterised by a number of waves P, QRS, T which are correlated to the status of the heart activity. In this paper, the aim is to present a powerful algorithm to aid cardiac diagnosis. The approach used is based on a determinist method, that of the tree decision. However, the different waves of the ECG signal need to be identified and then measured following a signal to noise enhancement. Signal to noise enhancement is performed by a combiner linear adaptive filter whereas P, QRS, T wave identification and measurement are performed by a derivative approach. Results obtained on simulated and real ECG signals are shown to be highly, satisfactory in the aid of cardiac arrhythmia diagnosis, such as junctionnal escapes, blocks, etc.