Automatic Detection of Characteristic Waves in Electrocardiogram

Lucia Billeci1, Lorenzo Bachi2, Maurizio Varanini3
1Institute of Clinical Physiology, National Research Council of Italy (CNR), 2Institute of Clinical Physiology, National Research Council of Italy (IFC-CNR), 3Institute of Clinical Physiology, National Research Council (CNR)


The goal of automatic ECG analysis is to assess the clinical status of the heart system as accurately as possible, and the identification of P and T waves plays a significant role in this matter. This works presents an original algorithm for the detection of P and T waves peaks. Pre-processing comprises a notch filtering stage to remove power-line interference if detected, baseline wandering removal by subtraction of a line connecting successive QRS onsets and a lowpass filtering stage implemented through a cascade of moving averages to cut off movement artifacts and residual noise. The filtered signal is then upsampled to 1 KHz. Since the ability to detect P and T waves critically depends on the correct positioning of R peaks, our algorithm was developed considering annotated R peaks. P-wave search algorithm was based on the search of the maximum area between the signal and a line, hinged to the signal in two points 0.080 s apart, running backwards from the point Q_on until the other end reaches an estimated position of the T wave in the previous complex. T-wave peaks are identified considering the sequence of the changes of the slope and the signal’s local extreme in an interval ranging from the S-wave offset to the time identified by Bazett’s formula.
To test and compare the algorithm’s performance, we considered the QTDB and MIT-BIH Arrhythmia annotated databases. On the QTDB database our algorithm obtained considerably higher performance than those presented in the literature (Friganovic et al., 2018) for both P and T wave (P: 94.83% vs 89.05%; T: 89.05% vs 87.49% for channel1), while on the MIT-BITH database the results were almost comparable to those reported in the literature. These findings suggest the high potential of the proposed simple algorithm for P and T wave detection in ECG.