Arrhythmic Electrocardiogram Beat Detection and Classification Using Heart Rate Variability Analysis

We present a system for automatic arrhythmia detection and classification in ECGs. Arrhythmia detection is based on time and timefrequency domain features of heart rate variability. The RR interval duration signal is extracted from ECGs and segmented into small intervals. Time domain features are extracted and used for the training of a set of neural networks. Short time Fourier transform, and several time-frequency distributions are used in the time-frequency analysis and the obtained features are used for the training of a set of neural networks. The arrhythmia classification is based on a set of rules. The rules use information from the RR interval duration signal and they are applied on a 3 RR interval sliding window. The proposed approach is evaluated using the MIT-BIH arrhythmia database. The results obtained are satisfactory for both sensitivity and specificity for arrhythmia detection. The detected beats are used for testing the classification approach and the results are presented.