New improved methodology for ECG signal compression

ECG of a person is being recorded for diagnosis of heart diseases, regular checkups, fitness and many other diseases also. Hence, huge amount of ECG data is being generated daily in hospitals. ECG recording and monitoring consume lots of memory space of digital computers. Data compression plays a vital role in reducing storage space and utilizing transmission bandwidth effectively. Objective of the research work is to propose a robust and effective method for ECG signal compression. This paper includes extraction of key morphological features, statistical features from decomposed signal, analysis in wavelet domain and classification of feature set. To reduce dimensionality of feature set, principle component analysis (PCA) is applied. Accuracy achieved with 15 principle components is same as pure wavelet transform with significant improvement in compression ratio (CR) by a factor of 4:1. MITBIH Arrhythmia database is used for experimentation. Accuracy obtained with K nearest neighborhood (KNN) classifier is 82.60% and sensitivity 92.30%. Research work carried out here is an improvement in the existing telemedicine technology.