Abstract:
The irregularity of heart beat is known as arrhythmia. This disease is sometimes very dangerous for a human body. It can cause death. So, proper treatment is a must in this case. Before treatment, proper diagnosis is needed. There are different types of arrhythmias. Treatment is different for different types. So, efficiently and correctly classification of ECG arrhythmias is a great challenge for treatment of this disease. There are a lot of techniques to classify the ECG signal. Some of them are very efficient, but the process is complicated. Again, some of them are easy and simple, but not very efficient. So, it is a great challenge for the researchers to find out a simple but efficient process for the classification. Our proposed method is quite simple and it is a general method for classifying ECG data. This method is based on extracting the features of ECG signal and classifying according to them. The efficiency of this method is quite good. We have used the MIT-BIH database for testing our method. This database is considered as the standard database in the whole world. We have trained some data for different types of arrhythmias and then taking those data as reference data, we tested other data and found out different features of the ECG signal. According to these features, we have classified Normal, Right Bundle Branch Block and Left Bundle Branch Block ECG signal. There are many other types of abnormal signal. In the future we will try to classify them. We have shown a comparison at the last of this thesis paper. From the comparison, we see that the efficiency of our method is good, but not better than most other methods. There are many scopes to improve our proposed method. So, we will try to improve the efficiency of this proposed method.
Description:
Supervised by
Prof. Dr. Mohammad Rakibul Islam,
Department of Electrical and Electronic Engineering(EEE),
Islamic University of Technology (IUT)
