ECG classification and abnormality detection using cascade forward neural network

Electrical activity of the heart is called as electrocardiogram i.e. ECG. Arrhythmias are among the most common ECG abnormalities. ECGs provide lots f information about heart abnormalities. The diagnosis depends upon the physician and it varies from physician to physician and also depends upon the experience of the physician. Previously many techniques were tried for analysis and automisation of the analysis. This paper describes the use of MATLAB based artificial neural network tools for ECG analysis for finding out whether the ECG is normal or abnormal and if it is abnormal, what is the abnormality. There are various arrhythmia like Ventricular premature beats, asystole, couplet, bigeminy, fusion beats etc. To classify this, various weighted neural networks were tried with different algorithms. They were provided training inputs from the standard MIT-BIH Arrhythmia database and tested by providing unknown patient data from the same database. The results obtained with different networks and different algorithms are compared, it is found that to identify whether the ECG beat is normal or abnormal, cascade forward back network algorithm has shown 99.9 % correct classification. These results are compared with previous neural network techniques and found that method proposed in this paper gives best results.