Cardiac arrhythmia detection using instantaneous frequency estimation of ECG signals

Preliminary detection and classification of cardiac arrhythmia is one of the most important problems in biomedical signal processing. This research proposes the estimation of instantaneous frequency (IF) of an ECG signal as a method for carrying out detection of cardiac disorder. Based on IF estimates, a classifier has been designed to differentiate a diseased signal from a normal one. Training, testing and validation of the classifier has been carried out using signals from MIT Arrhythmia, Normal Sinus Rhythm and Ventricular Arrhythmia databases, respectively. The sensitivity and specificity of the classifier comes out to be 97.82% and 100%, respectively and it is considered especially suitable for ambulatory ECG analysis.