Electrocardiogram Signal Data Processing

An electrocardiogram shows electrical activity of the heart. Doctors are used it to find out any abnormalities in the working of the heart. Manual inspection of ECG is very difficult as the signals of ECG are transient in nature. It is difficult to observe them within short span of time. Now a days most of the manual activities are computerized. So now a days computerized methods are used. These methods are able to tell the type of abnormality. These methods use the data produces by ECG. So the results of computerized inspection are totally dependent on the data produced by ECG. If the produced data is accurate then the result will be accurate. The data produced by ECO suffers from two drawbacks-first is baseline wander and second is noise. Both should be removed for perfect analysis of ECG. In this paper, a framework is produced for getting perfect ECG data. Baseline wander is removed by applying savitzky golay filter and noise is removed by using db6 wavelet upto level 5 decomposition. Then only approximation co-efficients are used for classification of ECG signals. It is found that the accuracy of classification of ECG signals reached to 100%.