Open Access

Sebastian Zaunseder, Wolf-Joachim Fischer, Rüdiger Poll, Matthias Rabenau

• This paper presents a wavelet-based signal processing method developed for an ambulatory ECG monitoring system. The monitoring system comprises modern trends in ambulatory ECG monitoring like integration of hardware in clothing, the use of low-power components and wireless data transmission via Bluetooth. The signal processing is located close to the sensor, thus allowing increased variability for the subsequent data handling (i.e. data transmission in case of detected abnormalities). Due to the very limited computational resources (an ultra-low power microncontroller (µC)) and the relatively high demands upon signal processing, the need arises for a method which meets the special demands of the ambulatory application. Therefore, we developed a wavelet-based method for detecting QRS complexes, especially adapted to the real-time requirements. The novel idea of our approach was to incorporate information gained from a lower scale directly into the threshold applied for QRS detection i nThis paper presents a wavelet-based signal processing method developed for an ambulatory ECG monitoring system. The monitoring system comprises modern trends in ambulatory ECG monitoring like integration of hardware in clothing, the use of low-power components and wireless data transmission via Bluetooth. The signal processing is located close to the sensor, thus allowing increased variability for the subsequent data handling (i.e. data transmission in case of detected abnormalities). Due to the very limited computational resources (an ultra-low power microncontroller (µC)) and the relatively high demands upon signal processing, the need arises for a method which meets the special demands of the ambulatory application. Therefore, we developed a wavelet-based method for detecting QRS complexes, especially adapted to the real-time requirements. The novel idea of our approach was to incorporate information gained from a lower scale directly into the threshold applied for QRS detection i n a higher scale. To date, all tests proved a very low computational load while simultaneously preserving the reliability of the analysis (Se=99,74%, +P=99,85% using the entire MIT-BIH Arrhythmia Database), thus pointing out the possibilities of real-time signal processing under ultra-low power conditions.…