Introduction: Cardiac arrhythmia, especially in the early hours, can stop heart activity and cause death. Since mistake in decision making is one of the most important causes of death in patients in cardiac intensive care units, the identification and classification of cardiac arrhythmias using the ECG signal is a valuable information source for diagnosing patients with heart abnormalities in times of crisis and warfare.
Method: At first, the noises of ECG signal are removed using digital filters and discrete wavelet transform (DWT). Then, Kurtogram of each QRS complex is obtained using spectral kurtosis analysis. Informative features are obtained from segments of Kurtogram function. Finally, K-nearest neighbor classifier is used to determine the normality of person or its arrhythmia type is detected. 
Results: In this paper, ECG signals from MIT-BIH signal are used. ECG signals of normal persons and four arrhythmias including APB, PVC, LBBB, and RBBB are chosen for classification. Obtained results show that proposed method achieves the accuracy of 98.51% for classification of ECG signals. Since accuracy of cardiac arrhythmia detection is an important and vital issue in medicine, the proposed method can be used by cardiologists to make a robust decision.
Conclusion: Considering the low computational complexity of the proposed method and obtained results, it can be used for fast and accurate cardiac arrhythmia detection, which is a special care resource and important task in physician within war.