Wave sequence based identification of sinus rhythm beats on a microcontroller

Holter recorders are currently changing their typical arrhythmia detection focus towards additional ECG evaluation objectives like ST-T-segment or heart rate variability analysis. Such estimations are only applicable when they are related to normal sinus rhythm excitations. However, up to now most approaches do not actively inquire this question but label every beat normal which is not sufficiently pathologic to be identified as abnormal beat. In this work we propose a real time applicable algorithm to identify sinus rhythm beats depending on their characteristic wave sequence regularity. Identification results are evaluated against the AAMI standard conform beat reference annotations in the MIT-BIH Arrhythmia database (Se=93.52%; +P=90.24%), European ST-T-Database (Se=95.09%; +P=99.86%) and the MIT-BIH Normal Sinus Rhythm database (Se=98.56%; +P=99.65%). Additionally we prove the algorithms to be running on an ARM Cortex-M3 microprocessor by detailed execution time and memory usage evaluation. The presented real time applicable algorithm allows an active beat by beat identification of sinus rhythm excitations to continue with comprehensive evaluations which rely on physiological conduction properties.