Paper Submission & Registration
8th Dutch Bio-Medical Engineering Conference
Go-previous
12:30   Heart - I
Chair: Elisabeth Wilhelm
12:30
15 mins
Temporal influence in ECG parameters for myocardial infarction detection
Alfonso Aranda Hernandez, Pietro Bonizzi, Joel Karel, Ralf Peeters
Abstract: Background Several studies have evaluated the use of different electrocardiographic (ECG) parameters to diagnose myocardial infarction (MI) in patients. Nonetheless, none of those have looked into the relation between ECG parameter values and time elapsed from MI onset to ECG recording. MI is a dynamic process and the ECG parameters used to characterize it can be influenced by these dynamics. For example, ST elevation increases in the acute phase of MI (from MI onset to several hours afterwards) and goes back to normal in the following days. In this study, we investigated the effect of time (from MI onset to ECG recording) on the values of ECG parameters used in the diagnosis of MI. Methods We used the PhysioNet PTB database as data source. This database contains 15 simultaneously recorded cardiac signals (12-lead ECG and three Frank orthogonal leads) from 290 subjects. The dataset includes healthy subjects and patients affected by different cardiac pathologies, among which 148 are MI patients. For the MI patients, the date of the ECG recording and infarction are available. We used this subset of patients in order to analyze ECG-based MI parameters change during the hours/days after infarction. The ECG based parameters considered are ST elevation and vectorcardiography (VCG) derived parameters. We split the patients in 5 groups depending on the amount of time passed between MI onset and ECG recording. Results We observed that ECG parameters seem to change based on the time between infarction and ECG recording time. The changes in the 5 time groups, comparing all possible pairs (10 in total), were statistically significant (p < 0.001). The trends were either ascending or descending depending on the ECG parameter considered. Conclusion Our study shows that in order to assess MI condition using ECG parameters, time from infarction to ECG recording is an important factor. This suggests that including such a factor in the diagnostic work up of MI patients could help in the automatic detection and characterization of MI.
12:45
15 mins
Model-based characterization of de novo POAF and persistent AF using 12-lead ECG signals
Hanie Moghaddasi, Borbála Hunyadi, Alle-Jan van der Veen, Natasja M.S. de Groot, Richard C. Hendriks
Abstract: Atrial fibrillation (AF) is the most common and sustained arrhythmia. AF is described by uncoordinated atrial activity that is represented on the electrocardiogram (ECG) by the irregular RR intervals and by fibrillatory waves or the absence of a P wave, instead of a single P wave. However, AF is also the most common post-cardiac surgical complication in patients without a history of AF. It is important to distinguish between AF with different durations, so short- and long-lasting AF episodes. For short-lasting AF, post-operative AF(POAF) is a good model, whereas patients already having pre-operatively AF are good representatives of long-lasting AF. Based on visual inspection of ECG signals with AF episodes, it is hard to find a difference between these two groups of patients. Since de novo POAF is the first time that a patient develops AF, comparing its evolution to persistent AF and being able to differentiate the two might lead to insights into the severity of the electrical changes. Moreover, being able to differentiate de novo POAF and persistent AF could give insights on how to determine the development stage of AF from the ECG. This knowledge is critical, as early and accurate detection will increase the chances of successful treatment (e.g., ablation therapy). In this work, we propose an automated method to discriminate between the characteristics of the various stages (durations) of AF from a multi-channel ECG. Based on a signal model of AF, we extract features that express the differences between two groups in terms of 1) rhythmic changes according to surveying variation of RR intervals and, 2) beat-to-beat variations in morphology and frequency components between the groups. Based on this, three sets of features are developed, including characteristics of the Poincare plots, vectorcardiogram, and normalized dominant frequency. Our classification system consists of a random forest, after a feature selection stage using the ReliefF method. The detection efficiency is validated on 151 patients and we achieved 87.3% accuracy. The results show that the features are useful to discriminate between the groups. Future work will aim at giving a physiological interpretation to unveil the underlying electropathology of AF.
13:00
15 mins
The impact of different regularization parameter values in electrocardiographic imaging
Tiantian Wang, Joel Karel, Pietro Bonizzi, Ralf Peeters
Abstract: Objectives: Electrocardiographic imaging (ECGI) reconstructs heart potentials from body-surface potentials by solving an ill-posed inverse problem requiring regularization. This study assesses the impact of the regularization parameter lambda on the quality of the inverse solution reconstructed with zeroth-order Tikhonov regularization. Data: Body-surface (184-216 electrodes) and epicardial potentials (99 electrodes) were recorded simultaneously in four healthy anesthetized dogs. In total, 92 different beats were analysed, including 6 sinus beats and 86 paced beats. Methods: The regularization parameter lambda was varied from 0.001-1.0, and reconstruction performed for all 92 beats. When comparing reconstructed with recorded epicardial electrograms, the correlation coefficient (CC) was computed after alignment, and relative error (RE) was calculated after normalization. We also compared activation times (AT) and recovery times (RT), in terms of CC and RE. Moreover, AT and RT isochrone maps were analysed for one sinus beat and three paced beats. We distinguished three regimes: (1) (low regularization) lambda in the range 0.001-0.009; (2) (medium regularization) lambda in the range 0.01-0.09; (3) (high regularization) lambda in the range 0.1-1.0. Results: Highest CCs and lowest REs between recorded and reconstructed electrograms all occur for lambda in the range 0.02-0.1. For AT, the CC is highest for lambda in 0.02-0.04, and for RT the CC is highest for lambda in 0.03-0.04. For AT, the RE is lowest for lambda in 0.009-0.06, and for RT the RE is lowest for lambda in 0.02-0.03. The isochrone maps for the sinus beat show a physiological and stable pattern for lambda in the range 0.02-0.09. For an LVant paced beat, activation isochrones show a physiological pattern for lambda in 0.03-0.04, and recovery isochrones for lambda in 0.02-0.04. For an LVlat paced beat, activation isochrones show a physiological pattern for lambda in 0.01-0.02, and recovery isochrones for lambda in 0.03-0.05. For an RVlat paced beat, activation and recovery isochrones show a physiological pattern for lambda in 0.009-0.05. Conclusions: Reliable ECGI consistently needs medium regularization, with lambda in the range 0.02-0.05. This analysis opens ways to assess performance of different regularization methods, and to investigate the influence of lambda in generating artefacts in the inverse solution.
13:15
15 mins
Fragmented QRS dynamics towards electrical storm in ICD patients
Amalia Villa, Sebastian Ingelaere, Sabine Van Huffel, Rik Willems, Carolina Varon
Abstract: Electrical storm (ES) in ICD patients, defined as 3 or more appropriate ICD interventions within a time span of 24 hours, is a medical emergency associated with adverse outcome [1]. However, it is debated if ES is only a marker of progressive near end-stage cardiac disease or an arrhythmogenic entity on its own. Better understanding and prediction are necessary to manage the burden of ES. Given the fact that arrhythmia is the presenting clinical picture in ES, current tools for arrhythmia prediction can be used to develop a better model. One of the most known markers of arrhythmia is the presence and level of fragmentation in the QRS complexes [2]. The goal of this study is to explore the relation between the presence of fragmented QRS (fQRS) and the manifestation of electrical storm in patients with an ICD for ischemic heart disease. From the UZ Leuven hospital ICD registry, 50 patients (94% male, 63±10 years, 37.3±14.3% LVEF at implant) with ischemic cardiomyopathy presenting with ES were identified. Baseline demographics and raw 12-lead ECG data were collected for these patients from implant until storm. As control subjects, 50 patients (94% male, 62±12 years, 34.2±12.2\% LVEF) from the same registry were included. 4 visits were considered for each ES patient: (1) from 36 to 18 months pre-ES, (2) from 18 to 6 months pre-ES, (3) from 6 months before to days before the storm and (4) from 0 to 48 hours after the storm. 4 visits similarly spaced in time were considered for the control group. Each of these visits consisted of 10 seconds of 12-lead ECG recordings, sampled at 250, 500 or 1000 Hz. The fQRS level in the 12-lead ECG data recorded in each visit was automatically quantified with a score between 0 and 1 for each lead, based on the method proposed in [3]. This method extracts features from the signal in the time domain, from Variational Mode Decomposition (VMD), and Phase Rectified Signal Averaging (PRSA), which are fed into a model trained from the annotations of 5 experienced cardiologists in another dataset presented in [3]. A Friedman test between the first and last visit for each of the groups showed a significant increase in the average level of fragmentation for the patients presenting ES, absent in the control group. This suggests that there is a trend towards deterioration in fQRS for patients manifesting ES with an ICD for ischemic heart disease. Further analysis will study the progress of these scores over the 4 visits, and will aim to integrate the fQRS scores of each independent lead our of the 12 set-up into a global score more interpretable.


end %-->