Arrhythmia Classification of ECG Signals Using Undecimated Discrete Wavelet Transform

Accurate arrhythmia classification of Electrocardiogram (ECG) signals in the initial stages is essential for effective diagnosis. It helps in the prevention of severe cardiac conditions. This has led to extensive research in the development of algorithms for heartbeat detection and real-time cardiac monitoring in recent years. This paper proposes an algorithm for the software analysis and arrhythmia classification of ECG signals. Each signal was decomposed using an Undecimated Discrete Wavelet Transform (UDWT). Further, an approximate version of the ECG waveform was reconstructed using the desired frequency bands, and followed by R-peak detection. Heart rate was estimated for the classification of types of arrhythmias. The UDWT-based technique is advantageous due to its ability to decompose signals at several levels, which can be utilized for various processing purposes. Additionally, it facilitates the simultaneous location of both time and frequency domains. The project work was implemented using MATLAB and SIMULINK. The data records from MIT-BIH and ECG-ID databases were used to test the algorithm.