Abstract:
The Electrocardiogram signals for different diseases comprises many indiscriminate features which makes the classification, in to a big task for diagnosis and treatment. The conventional method of classification has many inconvenience to the physicians that forces the public to look after the experienced cardiologist for their heart related diagnosis and treatment. So the situation made us to develop a new classifier to reduce high mortality rate of heart diseases, early prediction and precise classification of ECG signal. In this research the two important tasks needed for development of classifier which comes under data mining are clustering and classification. The data required for classification are extracted with Wavelet Transform (WT) and pre-classification were done with Fuzzy C-Means Clustering (FCM) technique. Finally the classifiers are developed with Feed Forward Neural Network (FFNN) by using the data, extracted by FCM. The simulated waveforms are used in training and testing and to classify three different arrhythmias. The techniques adapted in the design of classifier performs relatively well in terms of classification results, when compared with other classification technique.
