G. Lenis, N. Pilia, T. Oesterlein, A. Luik, C. Schmitt, and O. Dössel. P wave detection and delineation in the ECG based on the phase free stationary wavelet transform and using intracardiac atrial electrograms as reference. In Biomedizinische Technik. Biomedical Engineering, vol. 61(1) , pp. 37-56, 2016
Robust and exact automatic P wave detection and delineation in the electrocardiogram (ECG) is still an interesting but challenging research topic. The early prognosis of cardiac afflictions such as atrial fibrillation and the response of a patient to a given treatment is believed to improve if the P wave is carefully analyzed during sinus rhythm. Manual annotation of the signals is a tedious and subjective task. Its correctness depends on the experience of the annotator, quality of the signal, and ECG lead. In this work, we present a wavelet-based algorithm to detect and delineate P waves in individual ECG leads. We evaluated a large group of commonly used wavelets and frequency bands (wavelet levels) and introduced a special phase free wavelet transformation. The local extrema of the transformed signals are directly related to the delineating points of the P wave. First, the algorithm was studied using synthetic signals. Then, the optimal parameter configuration was found using intracardiac electrograms and surface ECGs measured simultaneously. The reverse biorthogonal wavelet 3.3 was found to be optimal for this application. In the end, the method was validated using the QT database from PhysioNet. We showed that the algorithm works more accurately and more robustly than other methods presented in literature. The validation study delivered an average delineation error of the P wave onset of -0.32+/-12.41 ms when compared to manual annotations. In conclusion, the algorithm is suitable for handling varying P wave shapes and low signal-to-noise ratios.
Student Theses (1)
N. Pilia. Characterization and reconstruction of ionic concentrations in the human ventricles analyzing the action potential and the surface ECG. Institute of Biomedical Engineering, Karlsruhe Institute of Technology (KIT). Bachelorarbeit. 2016
In the United States of America, 13.1% of the population suffer from chronic kidney dis- ease (CKD) . Patients suffering from end-stage CKD are often treated by haemodialysis where ionic concentrations of calcium, potassium and sodium get corrected. However, the frequency of deaths caused by cardiovascular diseases is up to 10% to 30% higher in patients undergoing dialysis than in the general population . The electrocardiogram (ECG) as a continuous, non-invasive monitoring tool could permit new insights into pos- sible links between heart diseases and changed ionic concentrations. In this work, methods for reconstruction of the ionic concentrations from the ECG are evaluated. In a first step, monodomain simulations with the ten Tusscher ventricular cell model  were performed on single cell, strand and whole heart level for different extracellular concentrations. The latter simulations were forward calculated and a standard 12-lead ECG was extracted. Extracellular sodium was varied between 120mmol/l and 150mmol/l, extracellular cal- cium between 0.6mmol/l and 3mmol/l, and extracellular potassium between 3mmol/l and 9 mmol/l as done in . These ranges are comparable to clinically observed ionic con- centrations. Sodium concentration changes showed minimal variations on all three simula- tion levels and were excluded from further investigations. Calcium and potassium changes yielded action potentials and ECGs clearly differing in amplitudes and morphologies. In a second step, 91 simulated ECG signals at different potassium and calcium concentrations - each including one beat - were used for reconstructing the ionic concentrations directly from the ECG. Features were extracted from the signals designed to describe changes caused by varied ionic concentrations. The inverse problem, i.e. coming back from the ECG features to the ionic concentrations was solved by regression. Two methods from literature  , linear regression, linear regression with regularisation (both with a linear and third order polynomial model), random forest regression and regression by artificial neural networks were implemented. Best results for potassium estimation were achieved by neural network regression yielding an estimation error of 0.007 mmol/l±0.2402 mmol/l (mean±standard deviation). Calcium estimation was best performed by third order poly- nomial regression achieving an estimation error of -0.0010 mmol/l±0.0511 mmol/l. Errors were calculated by 13-fold cross validation. Further investigations on the behaviour of the methods could be performed using simulated signals, which are generated with other models, and real measurements. Nevertheless, these first results seem to be very promis- ing: an adequate estimation of potassium and calcium concentrations could be shown to be possible.