Program Module for Encoding of QRS-Complexes Based on Morphological Features

In this work we made the program module for QRS-complexes encoding based on morphological features. This proposed program module allows to perform three processes: (1) the signal preprocessing of ECG signal, (2) QRS-complexes detections and (3) calculation of morphological features. In the preprocessing step we used the morphological filtering to remove isoline drift and noise with minimum source signal distortion. Multiscale morphological derivative detector was used for QRS-complexes detection. After QRS-complexes detection morphological features were calculated. This proposed program module is developed using Matlab 2008a Mathwork, Inc. This program module was tested using the MLII ECG leads from the MIT-BIH arrhythmia database. We tested this module to detect QRS-complexes for 5 types of heartbeats: Normal beat (NB), Left Bundle Branch Block (LBBB) Beat, Right Bundle Branch Block (RBBB) beat, Atrial Premature Contraction (APC) beat and Premature Ventricular Contraction (PVC) beat. In this work experimental results showed that the accuracies of QRS-complexes detection for above 5 types of heartbeats are as follows: NB - 92 %, LBBB - 98 %, RBBB - 94%, AVC - 97% and PVC - 80%. The computed values of morphological characteristics can be used in encoding of ECG signals for classification of heartbeats and assignment of cardiac diagnosis.