| Automatic estimation of heart rates (HR) from ECG data is an important technique used in diagnosis of various heart diseases. Whether the HR estimate is reliable is the premise for obtaining the accurate results.;Methods for quantifying the quality of ECG data for the purpose of automatic estimation of heart rates are exiguous. We propose two algorithms, SAECG and K-means reliability evaluation methods, for evaluating the reliability of HR estimates. Both methods compare PQRST segments to a pre-computed template using three distance measures including PRO, CCORR and WDIST resulting in reliability indices. We use three types of noise including baseline wander, EMG artifacts and motion artifact to contaminate clean ECG signals for testing. Results show that the reliability values correlate well with HR estimation errors. Both methods exhibit the ability to track the accuracy of HR estimation results, i.e. they can effectively reflect the quality of the ECG signals. |
