Abstract:
Atrial Fibrillation (AF) is one of the most common cardiac arrhythmias. The number of patients related to heart failure due to AF is increasing day by day. Early detection of AF may reduce the risk of death due to heart failure. So, it has become more important to detect AF. There are various method to detect AF. In this thesis, we use ECG signal for AF detection. The MIT-BIH Atrial Fibrillation database is used to import ECG data for analysis. Filtered ECG signal using multistage multirate system for removing noise. RR interval of the ECG signal is calculated. Here we use the algorithm that mainly follows statistical method for detection of AF. Parametric statistic RMSSD and SE, and non-parametric statistic, TPR are used for this purpose. MATLAB R2016a is used to measure the values of those parameters for estimation of AF. The threshold values of RMSSD/ (Mean RR) taken from the literature is 0.1, SE is 0.7 and TPR is greater than 0.54 and lesser than 0.77. The resultant values of RMSSD, SE and TPR of every beat are checked weather it crosses the threshold level or not. If all the three parameters cross the threshold level then the beat flagged as AF. It shows excellent result when compared with the annotations of the database, and then the sensitivity, specificity and accuracy are determined. The algorithm has the sensitivity of 98.03%, specificity of 98.80% and accuracy of 99.45%. Thus, the result obtained in this study is appreciable compared to the other study found in literature.
Description:
This thesis is submitted to the Department of Electronics and Communication Engineering, Khulna University of Engineering & Technology in partial fulfillment of the requirements for the degree of
Master of Engineering (M.Sc. Eng.) in Electronics and Communication Engineering, November 2018.
Cataloged from PDF Version of Thesis.
Includes bibliographical references in each chapter.
