Abstract:

This study examines the effect of mental workload on the electrocardiogram (ECG) of participants driving the Lane Change Task (LCT). Different levels of mental workload were induced by a secondary task (n-back task) with three levels of difficulty. Subjective data showed a significant increase of the experienced workload over all three levels. An exploratory approach was chosen to extract a large number of rhythmical and morphological features from the ECG signal thereby identifying those which differentiated best between the levels of mental workload. No single rhythmical or morphological feature was able to differentiate between all three levels. A group of parameters were extracted which were at least able to discriminate between two levels. For future research, a combination of features is recommended to achieve best diagnosticity for different levels of mental workload.

Abstract:

Heart rate variability (HRV) plays an important role in medicine and psychology because it is used to quantify imbalances of the autonomic nervous system (ANS). An important manifestations of the ANS on HRV is also directly related to respiration and it is called respiratory sinus arrhythmia (RSA). This is a controlled phenomenon that leads to a synchronized coupling between respiration and instantaneous heart rate. Thus, the portion of HRV that is not related to respiration, and could potentially contain undiscovered diagnostic value, is overlapped and remains hidden in a standard HRV analysis. In such cases, a decoupling procedure would deliver a discriminated HRV analysis and possible new insights about the regulation of the cardiovascular system. In this work, we propose an algorithm based on Granger's causality to measure coupling between respiration and HRV. In the case of significant coupling, we estimate and cancel the respiration driven HRV component using a linear filtering approach. We tested the method using synthetic signals and prove it to deliver a reliable coupling measurement in 96.3% of the cases and reconstruct respiration free signals with a median correlation coefficient of 0.992. Afterwards, we applied our method to signals recorded during paced respiration and during natural breathing. We demonstrated that coupling is dependent on respiratory frequency and that it maximizes at 0.3 Hz. Furthermore, the HRV parameters measured during paced respiration tend to level among subjects after decoupling. The intersubject variability of HRV parameter is also decreased after the separation process. During natural breathing, coupling is notoriously lower to non-existing and decoupling has little impact on HRV. We conclude that the method proposed here can be used to investigate the diagnostic value of respiration independent HRV parameters.

Abstract:

The most important ECG marker for the diagnosis of ischemia or infarction is a change in the ST segment. Baseline wander is a typical artifact that corrupts the recorded ECG and can hinder the correct diagnosis of such diseases. For the purpose of finding the best suited filter for the removal of baseline wander, the ground truth about the ST change prior to the corrupting artifact and the subsequent filtering process is needed. In order to create the desired reference, we used a large simulation study that allowed us to represent the ischemic heart at a multiscale level from the cardiac myocyte to the surface ECG. We also created a realistic model of baseline wander to evaluate five filtering techniques commonly used in literature. In the simulation study, we included a total of 5.5 million signals coming from 765 electrophysiological setups. We found that the best performing method was the wavelet-based baseline cancellation. However, for medical applications, the Butterworth high-pass filter is the better choice because it is computationally cheap and almost as accurate. Even though all methods modify the ST segment up to some extent, they were all proved to be better than leaving baseline wander unfiltered.

Abstract:

Today, patients suffering from atrial arrhythmias like atrial flutter (AFlut) or atrial fibrillation (AFib) are examined in the EP-lab (electrophysiology lab) in order to understand and treat the disease. Multichannel catheters are advanced into the atria in order to measureelectric signals at manyintracardiacpositions simultaneously. Complementary to clinical learning,comprehension of the disease and therapeutic strategies can be improved with computer modeling of the heart. This way, hypotheses about initiation and perpetuation of the arrhythmia can be tested and ablation strategies can be assessed in-silico. Modeling and biosignal analysis can benefit from mutual fertilization. On the one hand, modeling can be improved and personalization can be achieved via high density mapping of the atria. On the other hand, new algorithms for the interpretation of multichannel electrograms can be developed and evaluated with synthetic signals from computer models of the atria. This article illustrates the synergetic potential by examples and highlights challenges to be addressed in the future.

Abstract:

Cardiac arrhythmias are a widely spread disease in industrialized countries. A common clinical treatment for this disease is radiofrequency ablation (RFA), in which high frequency alternating current creates a lesion on the myocardium. However, the formation of the lesion is not entirely understood. To obtain more information about ablation lesions (ALs) and their electrophysiological properties, we established an in-vitro setup to record electrical activity of rat myocardium. Electrical activity is measured by a circular shaped multielectrode array. This work was focused to gain more information by developing algorithms to process the measured electrical signals to collect different features, which may allow us to characterize an AL. First, pacing artefacts were detected and blanked. Subsequently, data were filtered. Afterwards, activations in atrial signals were detected using a non-linear energy operator (NLEO) and templates of these activations were generated. Finally, we determined different features on each activation in order to evaluate changes of unipolar as well as bipolar electrograms and considered these features before and after ablation. In conclusion, the majority of the signal features delivered significant differences between normal tissue and lesion. Among others, a reduction in peak to peak amplitude and a diminished spectral power in the band 0 to 100 Hz may be useful indicators for AL. These criteria should be verified in future studies with the aim of estimating indirectly the formation of a lesion.

Abstract:

We examined if ECG-based features are discrimi-native towards drowsiness. Twenty-five volunteers (1932 years) completed 7×40 minutes of monotonous overnight driving simulation, designed to induce drowsiness. ECG (512 s-1) was recorded continuously; subjective ratings of drowsiness on the Karolinska sleepiness scale (KSS) were polled every five minutes. ECG recordings were divided into 5-min segments, each associated with the mean of one self- and two observer-KSS ratings. Those mean KSS values were binarized to obtain two classes not drowsy and drowsy. The Q-, R- and T-waves in the recordings were detected; R-peak positions were manually reviewed; the Q- and T-detection method was tested against the manual annotations of Physio-nets QT database. Power spectral densities of RR intervals (RR-PSD) and quantiles of the empirical distribution of heart-rate corrected QTc intervals were estimated. Support-vector machines and random-holdout cross-validation were used for the estimation of the classification error. Using either RR-PSD or QTc features yielded mean test errors of 79.3 ± 0.3 % and 82.7 ± 0.5 %, respectively. Merging RR and QTc features improved the accuracy to 88.3 ± 0.2 %. QTc intervals of the class drowsy were generally prolonged com-pared to not drowsy. Our findings indicate that the inclusion of QT intervals contribute to the discrimination of driver sleepiness.

Abstract:

Atrial arrhythmias such as atrial flutter and atrial fibrillation are a burden for patients and a major challenge for modern healthcare systems. Identification of patients at risk to develop atrial arrhythmias at an early stage carries the potential to reduce the incidence by implementing appropriate strategies to mitigate the risks. Diagnostic methods based on the ECG are ideal risk markers due to their noninvasiveness and omnipresence. The left atrium (LA) plays a major role in the intiation and perpetuation of atrial reentry arrhythmias. However, the LA is not well represented in the P-wave derived through standard ECG leads. Here, we optimize ECG lead positions to maximize LA information content. Towards this end, we used a cohort of eight personalized computational models providing the unique opportunity to separate LA and right atrial (RA) contributions to the P-wave, which is not feasible in vivo. The location of maximum P-wave signal energy was located on the center of the chest for all subjects with marked overlap between regions of maximum LA and RA P-wave amplitude. The regions of highest ratio between LA and RA signal energy differed between patients. However, a region with LA signal energy being higher than that of the RA and providing a sufficiently large absolute P-wave amplitude was identified at the center of the back consistently across five models of the cohort. Optimized linear combinations of standard 12-lead signals yielded comparably good results. Our newly proposed electrode positions on the back as well as selected linear combinations of standard 12-lead signals improve the LA information content considerably. By using these, more relevant diagnostic information regarding the anatomical and electrophysiological properties of the LA can be derived in future.

Abstract:

Cardiac excitation during atrial fibrillation (AFib) is changing dynamically, compromising the ability to identify underlying mechanisms by intracardiac catheter mapping. Statistical analysis of dominant excitation patterns may help to identify and subsequently eliminate the drivers of this tachycardia. As the morphology of local bipolar intracardiac electrograms (EGMs) depends on the orientation of the propagating excitation wave, its evaluation for a fixed multichannel catheter position can provide information about the stability of the depolarization pattern. Up to date, analysis of morphology is most often done by computing a similarity index or the recurrence rate of individual EGMs, reflecting how often similar excitations appear. We sougth to extend this approach to a classification based analysis technique. In each multichannel EGM, local activation waves (LAWs) were automatically detected by assessing instantaneous signal energy. A greedy algorithm was implemented to cluster LAWs based on their similiarity. New clusteres were formed when similarity fell below a predefined threshold. The concept was tested using simulated EGM data (quadratic patch of cardiac tissue, bidomain simulation, both planar and focal excitations, various catheter types). Results demonstrated that the algorithm correctly identified and classified the simulated excitation patterns. Subsequent quantitative analysis allowed to both discard singular classes of excitation and identify dominant excitations. The presented method forms the basis for statistical assessment of prevailing depolarization patterns, and for computation of additional features like conduction velocity, presence of focal sources, or dissociation when applied on multichannel data.

Abstract:

A common treatment of focal ventricular tachycardia is the catheter ablation of triggering sites. They have to be found manually by the physician during an intervention in a catheter lab. Thus, a method for determining the position of the focus automatically is desired. The inverse problem of electrocardiography addresses this problem by reconstructing the source of the ectopic beats using the surface ECG. This problem is ill-posed and therefore needs specific methods for solving it. We propose a machine learning approach for localisation of the ectopic foci in the heart to assist cardiologists with their therapy planning.We simulated 600 120-lead ECGs with different known excitation origins in the heart using a cellular automaton followed by a forward calculation. Features from the ECGs were used as input for a support vector regression (SVR). We assumed a functional relation between features from the ECG and the excitation origin. To benchmark SVR, we also used the well-known Tikhonov 0th order regularisation to reconstruct the transmembrane potentials in the heart and detect the location of the ectopic foci. Parameters for SVR and regularisation were chosen using a grid search minimising the error between estimated and true excitation origin. Compared to the Tikhonov regularisation method, SVR achieved a smaller deviation between estimated and real excitation origin evaluated with 6-fold cross validation. Future work could investigate on the behaviour on data from simulations with other torso and electrophysiological models, the influence of other methods for feature extraction and finally the evaluation with clinical data.

Abstract:

Intracardiac electrogram recordings during atrial fibrillation (AFib) are characterized by irregular rhythms and complex morphologies. Hence, analysis in the time domain is a difficult task. The so called dominant frequency DF is a spectrum based approach that aims at finding the most relevant frequency in a signal providing information about the rate and dynamics of AFib. However, in recent years various studies reported controversial results regarding the clinical relevance of the DF. In this work, a definition of the DF at a fundamental scale is proposed as the rate at which action potentials are triggered in atrial cells. The most common method to estimate the DF in literature, labeled as DFSpec, is examined in comparison to the proposed definition. A signal processing study using synthetic signals verified that the DFSpec is stable for all changes in morphology of atrial activations. However, it is also demonstrated that the DFSpec becomes unstable for variations above 20% in the cycle length of a signal. Spectrum based DF estimation should be interpreted in a critical manner and is not advisable for study endpoints or clinical markers.

Abstract:

The heart rate variability (HRV) is a measure which is commonly used to assess sympathetic and parasympathetic auto-nomic function. It is well known, that respiration can have a strong influence on HRV. Especially, a phenomenon called Respiratory Sinus Arrythmia (RSA) modulates the RR intervals and is a major contributor to the HRV. The interpreta-tion of common HRV parameters can be ambiguous due to different respiration rates and patterns. To assess this ambi-guity, the coupling of RSA on HRV was quantified and the HRV parameters were compared during different respirato-ry states.A pilot study with five healthy subjects was performed. A three lead ECG was acquired and the respiration was estimat-ed by measuring the aeration of the lungs using the PulmoVista 500 by Dräger. This device uses Electrical Impedance Tomography (EIT) to monitor impedance changes due to the changing amount of air within the lungs during respira-tion. The subjects were asked to breath at controlled respiration rates of 8, 15 and 24 breaths per minute as well as spon-taneously for 1 min each. In addition, to analyze HRV during apnoic phases without any respiration, the subjects were asked to hold their breath for 40s at end-inspiration and end-expiration. After preprocessing of the ECG and the respiration signal, the coupling between the measured respiration and the RR intervals was quantified using the Granger causality. If significant coupling was present, the HRV was separated from its respiratory influence using an ARMAX model. The measured respiration hereby formed the exogeneous input to the filter. Finally, common HRV parameters were calculated for the original and the decoupled RR intervals.We showed, that coupling strength depends on respiratory rates, which might complicate HRV interpretation. Moreo-ver, the coupling is decreased during spontaneous breathing in comparison to controlled respiration. Additionally we found, that HRV parameters during apnoic phases differ from decoupled HRV parameters during spontaneous or con-trolled respiration.

Abstract:

Atrial fibrillation and atrial flutter are the most common atrial arrhythmias placing a heavy burden on patients and posing a challenge on healthcare systems. If patients at risk to develop atrial arrhythmias can be identified at an early stage, the arrhythmia incidence can be lowered by implementing appropriate strategies to mitigate the risks. Diagnostic methods based on the ECG are ideal risk markers due to their noninvasiveness and omnipresence. The left atrium (LA) plays a major role in the initiation and perpetuation of atrial reentry arrhythmias. However, the LA is not well represented in the P-wave derived through standard ECG leads. Here, we optimize ECG leads to maximize LA information content. Towards this end, we used a cohort of eight personalized computational models providing the unique opportunity to separate LA and right atrial (RA) contributions to the P-wave, which is not feasible in vivo. The location of maximum P-wave signal energy was located on the center of the chest for all subjects with marked overlap between regions of maximum LA and RA P-wave amplitude. The regions of highest ratio between LA and RA signal energy differed between patients. However, a region with LA signal energy being higher than that of the RA and providing a sufficiently large absolute P-wave signal energy was identified at the lower left quadrant of the back consistently across most subjects of the cohort. Optimized linear combinations of standard 12-lead signals (considering the eight independent leads) yielded comparably good results amplifying LA information by more than one order of magnitude. Our newly proposed electrode positions on the back as well as selected combinations of standard ECG signals improve the LA information content considerably. By using these, more relevant diagnostic information regarding anatomical and electrophysiological properties of the LA can be derived in future.

Abstract:

Chronic kidney disease appears worldwide. In the United States, the number of patients suffering from kid- ney failure doubled from 1998 to 2010. A common treat- ment for these patients is haemodialysis. 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. To analyse the underly- ing effects and for a possible risk prediction, a continuous monitoring of the ionic concentrations that are influenced by dialysis is desired. In this work, a method for the re- construction of the ionic concentrations of calcium and potassium from the ECG is proposed. In a first step, 91 monodomain simulations with the ten Tusscher ventricular cell model were performed for different extracellular ionic concentrations. From there, a standard 12-lead ECG was extracted. Calcium and potassium changes yielded ECGs clearly differing in amplitude and morphology. In a second step, the simulated ECG signals were used for reconstruc- ting the ionic concentrations directly from the ECG. Fea- tures were extracted from the signals designed to describe changes caused by varied ionic concentrations. The in- verse problem, i.e. coming back from the ECG features to the ionic concentrations was solved by regression with an artificial neural network. Results for potassium estimation yield an error of 0.00±0.28 mmol/l (mean±standard de- viation) calculated with 7-fold cross validation. The esti- mation error for calcium was 0.00±0.08 mmol/l. Although these results underline the suitability of the method, the used ECGs differed from the observed in a clinical envi- ronment. However, simulations allow an evaluation un- der controlled conditions of a particular effect that was intended to be investigated. As the application to clinical data is yet missing, this study can be seen as a proof of concept showing that an artificial neural network is capa- ble of exactly estimating potassium and calcium concen- trations from ECG features. 1. Introduction Haemodialysis therapy is a common treatment method for patients suffering from chronic kidney disease (CKD) in the terminal stage. The amount of people in the United States suffering from kidney failure increased from 320,000 in 1998 to 650,000 in 2010. The frequency of deaths caused by cardiovascular events within the dialysis patient group is up to 10% to 30% higher than in gene- ral population [1]. Patients suffering from end-stage CKD experience high variations of blood electrolyte concentra- tions. These can directly influence the functioning of the heart. Thus, research on cardiovascular links could im- prove therapy and risk stratification. One tool which is capable of capturing the electrophysiological properties of the heart in a non-invasive way is the electrocardiogram (ECG). It is known, that electrolyte concentrations of po- tassium (K+) and calcium (Ca2+) affect the ECG [2]. Un- til now, a determination of the concentrations is connec- ted to a blood test. Hence, continuous monitoring of the ionic concentration is impracticable. However, the ECG as a continuous, non-invasive monitoring tool could shed a light on the relation between heart diseases and changes in the ionic concentration particularly after leaving the strictly supervised clinical area where dialysis takes place, i.e allowing a monitoring at home. Articles have been pub- lished showing that the reconstruction of extracellular K+ concentration can be done using just one feature from the ECG with a quadratic regression [3]. In this study, we tried to estimate both K+ and Ca2+ concentrations from the ECG. Therefore, we examined simulated ECGs at dif- ferent concentration levels and designed features descri- bing the observed changes in the ECG. A subset of these was used in connection with a machine learning method to reconstruct the concentrations. 2. Methods 2.1. Simulations A total number of 91 computer simulations of the car- diac electrophysiology were performed at whole heart

Abstract:

The electrocardiogram (ECG) captures the electrical activity of the heart that is projected onto the surface of the body. This signal can be recorded in a simple and cost effective manner making it available for a wide variety of mobile and stationary applications. Thus, over the last 100 years, the ECG has become the gold standard for the diagnosis of many cardiac afflictions. This is still relevant nowadays because cardiovascular diseases are a major topic of concern for our society accounting for almost 30 % of all causes of death worldwide. In particular, the ischemic heart disease is the single most common cause of death. Other cardiac arrhythmias, such as atrial fibrillation and atrial flutter, affect approximately 2 to 3 % of the population in the European Union leading to estimated costs of about 26 billion euros per year. In all these cases, the ECG is the mandatory first step leading to a reliable diagnosis and successful treatment.In this thesis, we have developed a series of signal processing algorithms capable of automatically extracting rhythmical and morphological properties from the ECG with the aim of supporting the decision making in the diagnostic process. In our first research project, we investigated a phenomenon called postextrasystolic T wave change (PEST) and postulated that the biomarkers obtained from the ECG during PEST could be used to predict pump failure progression death (PFD). The second project dealt with the creation of an algorithm to accurately detect and delineate the P wave in the ECG using as ground truth the electrograms recorded inside the atria. Our third investigation aimed at a deeper understanding of a physiological phenomenon called respiratory sinus arrhythmia (RSA). Here, we developed an algorithm that separates the heart rate variability (HRV) into a respiration driven component and a respiration independent part. The respiration free HRV could deliver new insights about the regulation of the cardiovascular system. In the fourth and final study, we investigated the impact of mental workload on the ECG while driving a car and discovered a variety of features that can help to detect a dangerous state of mind and protect the driver from a car crash.We conclude that well designed signal processing methods for the ECG have the potential of reducing the burden for the cardiac patient and the amount of accidents on the road