Morphology‐Enhanced Atrial Event Classification Improves Sensing in Pacemakers

Abstract

In atrial-based pacing, appropriate therapy and reliable diagnostics depend on detection and discrimination of atrial signals. Accurate classification of atrial events is mainly confounded by oversensing of ventricular far-field R-wave signals (FFRW), but attempts to reject FFRWs by manipulating atrial sensitivity and/or postventricular atrial blanking period (PVAB) may result in undersensing (especially of atrial fibrillation, AF) or in 2:1 atrial flutter detection. The objective of this study is therefore to evaluate if such methods can be improved by morphology-enhanced atrial event classification (MORPH). Twenty-four-hour ambulatory atrial electrograms were recorded from continuous telemetry of digital pacemakers. Half of the recording was used for collecting two individual morphology parameters that discriminated P-waves from FFRWs in every patient (learning phase). The other half was used to test the MORPH algorithm against traditional methods (classification phase). In 44/48 patients, data were suitable for analysis. Average P and FFRW amplitudes were 1.96 mV versus 0.61 mV (P < 0.001). The interval between ventricular events and FFRW oversensing (VA interval) averaged at 14 ms during sensing and at 118 ms during pacing in the ventricle. Compared to nominal ("Factory") settings, the MORPH algorithm improved the sensitivity for P-wave recognition from 97.2% to 99.2%, the specificity from 91.9% to 99.96%, and the accuracy from 95.3% to 99.4% (P < 0.01 for all). By improving atrial signal discrimination, morphology analysis of atrial electrograms allows for high atrial sensitivity settings, and potentially improves the reliability of atrial arrhythmia diagnostics in heart rhythm devices.