Rüdiger Poll

Optimization of ECG Classification by Means of Feature Selection

This study tackles the ECG classification problem by means of a methodology, which is able to enhance classification performance while simultaneously reducing the computational resources, making it specially adequate for its application in the improvement of ambulatory settings. For this purpose, the sequential forward floating search (SFFS) algorithm is applied with a new criterion function index based on linear discriminants. This criterion has been devised specifically to be a quality indicator in ECG arrhythmia classification. Based on this measure, a comprehensive feature set is analyzed with the SFFS algorithm, and the most suitable subset returned is additionally evaluated with a multilayer perceptron (MLP) to assess the robustness of the model. Aiming at obtaining meaningful estimates of the real-world performance and facilitating comparison with similar studies, the present contribution follows the Association for the Advancement of Medical Instrumentation standard EC57:1998 and the same interpatient division scheme used in several previous studies. Results show that by applying the proposed methods, the performance obtained in similar studies under the same constraints can be exceeded, while keeping the requirements suitable for ambulatory monitoring.

Prolonged Wearable ECG Monitoring - a Wavelet Based Approach

This paper presents the implementation of an ambulatory electrocardiogram (ECG) monitoring system. The proposed system comprises current trends on ambulatory ECG monitoring like integration of the hardware in clothing, the use of low power components, wireless data transmission via Bluetooth and the use of a PDA. Differing from other approaches the signal processing was located close to the sensor. Thus, more variability in further data handling and therefore a more efficient operation is possible. Employing adequate hardware components on the one hand and adapting a wavelet based signal processing method to the applied components on the other hand, we realized a system optimized in terms of suitability for daily use, energy efficiency and reliability.

A surface acoustic wave sensor for detection of cell adhesion

In this paper preliminary results for the detection of the cell adhesion process with shear-horizontal polarized surface acoustic wave devices (SH-SAW) are presented. The murine fibroblast cell line L929 was investigated. Therefore a special measurement setup was developed. It enables us to perform long term measurements at stable ambient conditions. Different amounts of cells where seeded. According to the number of cells different slopes of resonant frequency increase vs. time were found.

Combination of surface acoustic wave measurement and impedance spectroscopy for detection of cell adhesion process

In this study, a new way of investigating a surface acoustic wave (SAW) based sensor device is reported. For that, SAW sensors with a fundamental mode of about 85 MHz and a metallization of gold were developed. Realized as one port resonators, these sensors allow the continuous detection of electric and viscoelastic properties of the same living cell culture in one measurement step. For obtaining the electric properties of the cells the sensors response in a low frequency range up to 10 MHz (impedance spectroscopy) can be investigated. For the detection of viscoelastic properties, the change of the sensors resonant frequency and its impedance are monitored. All signals are detected vs. time and can be compared face to face.

Comparative Analysis of different Wigner-Ville Distribution Implementations for the ECG-based Detection of Obstructive Sleep Apnea

Obstructive sleep apnea (OSA) syndrome is a common breathing abnormality. During apnea, the airflow is impeded or totally interrupted. The reaction of the autonomic nervous system terminates the apnea and also leads to changes in heart rate variability (HRV). As shown in previous studies, the spectral analysis of HRV allows for a diagnosis of apnea. Therefore, a high quality time-frequency distribution is of great significance. The Wigner-Ville-Distribution (WVD) offers a very high resolution in both, time and frequency. However, the proper handling of cross terms resulting in the calculation of the WVD is a crucial point using the WVD. To cope with this task, the presented work compares different methods regarding their ability of cross term suppression and applicability. Furthermore it is shown that using spectral information of ECG overnight recordings from the Physionet Apnea Data-base, these datasets can be separated.

Real time recognition of complete atrioventricular blocks

Common heart arrhythmia monitors are limited to the electrical activity of the ventricles. Severe cardiac malfunctions exhibit typical defects of atrioventricular conduction which often can be efficiently treated with medicine or electrotherapy. The method described here enables the detection of complete atrioventricular blocks and may be used for the improvement of arrhythmia monitoring.