Arrhythmias detection and classification base on single beat ECG analysis

The effective manual detection ECG arrhythmia is very important, but it is tedious and time consume. Due to the ECG signal, monitoring may have to be carried out over several hours because the volume of the ECG data is enormous. This difficulty turns out a very high possibility of the analyst missing (or misreading) vital information. Therefore, computer-based analysis and detection of diseases can be very helpful in cardiologist's diagnoses. This paper proposes an algorithm to detect and classify the ECG arrhythmia, which is combined of the novel ECG beat length selection, Discrete Cosine Transform as the feature extraction, and Fisher's Linear Discriminant Analysis as the classifier system. The experimentation results demonstrate that the proposed algorithm classifies five arrhythmia types: normal, left bundle branch block, right bundle branch block, premature ventricular contraction, and atrial premature contraction beat. With the achievement results of 99.11% in terms of Total classification accuracy, 97.01% in terms of sensitivity, and 99.44% in terms of specificity. These obtained results are better than the other existing methods.