Application of Stockwell Transform and Shannon Energy for Pace Pulses Detection in a Single-Lead ECG Corrupted by EMG Artifacts

Electrocardiogram (ECG) analysis is important for the detection of pace pulse artifacts, since their existence indicates the presence of a pacemaker. ECG gives information on the proper functionality of the device and could help to evaluate the reaction of the heart. Beyond the challenges related to the diversity of ECG arrhythmias and pace pulses, the existence of electromyogram (EMG) noise could cause serious problems for the correct detection of pace pulses. This study reveals the potential of a methodology based on Stockwell transformation (S-transform), subsequent Shannon energy calculation and a threshold-based rule for pace artifact detection in a single-lead ECG corrupted with EMG noise. The design, validation and test are performed on a large, publicly available artificial database acquired with high amplitude and time resolution. It includes various combinations of ECG arrhythmias and pace pulses with different amplitudes, rising edges and total pulse durations, as well as timing that corresponds to different pacemaker modes. The training was done over 312 (ECG + EMG) signals. The method was validated on 390 clean ECGs and independently tested on 312 (ECG + EMG) and 390 clean ECGs. The achieved accuracy over the test dataset was Se = 100%, PPV = 98.0% for ECG corrupted by EMG artifacts and Se = 99.9%, PPV = 98.3% for clean ECG signals. This shows that, despite EMG artifacts, the S-transform could distinctly localize the pace pulse positions and, together with the applied ShE, could provide precise pace pulses detection in the time domain.

Electrocardiographic patterns of cardiac pacemakers: normal and abnormal findings

Pacemaker electrocardiograms (ECGs) have to be interpreted in the same systematic way as any other ECG tracing. In addition to the normal steps, however, an analysis of the pacing mode, of any arrhythmia involved, of pacemaker malfunction, and of dedicated pacemaker algorithms must be performed in a ‘step-up’ fashion from simple to complex. Some of these steps may require information about the pacemaker model and programmed parameters. Pacemaker malfunction consists of either pacing or sensing malfunction. Pacing malfunction occurs as ineffective pacing (i.e. loss of capture), either in the atrium or in the ventricle. In loss of capture, the pacing spike is visible in the ECG but is not followed by a P wave or QRS complex. Sensing malfunction consists of either under- or oversensing. In undersensing, intrinsic activity (P wave, QRS complex) is not detected and pacing is not inhibited. Therefore, the atrial or ventricular stimulus occurs earlier than expected, typically shortly after an intrinsic P wave or QRS complex. Oversensing is defined as sensing of unwanted signals by the atrial or ventricular lead. Sources of oversensing can be cardiac (e.g. far-field R-wave oversensing in the atrium, T-wave oversensing in the ventricle), other biological signals (e.g. pectoral or diaphragmatic myopotent oversensing), or electromagnetic interference (e.g. electrocautery, magnetic resonance imaging). Oversensing inhibits the scheduled pace and can therefore cause bradycardia and asystole. A systematic approach to pacemaker ECG is the basis for improving our diagnostic capabilities and improving the management of patients implanted with a cardiac device.