Computer Aided Atrial Fibrillation Detection from the Statistical Attributes of ECG Signal

Nowadays Atrial Fibrillation (AF) has become the most common heart arrhythmias. The AF-influenced patient's number is expanding day by day. The causes of heart failure can be decreased by recognizing the AF at an early stage. Electrocardiogram (ECG) is a non-invasive method for diagnosing a variety of common cardiac issues. A long ECG signal is difficult to read manually and an automatic computer-aided method for AF detection from the ECG signal is recommended. In this work, we present a beat-by-beat counting approach to detect AF. We use the statistical characteristics of the AF condition concerning the normal ECG characteristics. We perform baseline wander removal, noise cancellation, QRS detection, R-R interval measurement & P-wave amplitude measurement. With the statistical value of AF -affected ECG compared to the normal ECG, the proposed algorithm has achieved an accuracy of 77.91%. The outcomes of the proposed method create hope for computer aided AF detection widely.