A semi-automatic method for ECG analysis in left bundle branch area pacing

Background Assessing 12 lead ECGs during left bundle branch area pacing (LBBAP) is essential to determine whether the conduction system is captured. In current practice, ECG assessment is usually performed manually, either using calipers on the electrophysiology system or by measuring intervals on (printed) ECG grids. This process is time-consuming and subjective, and it may be insufficient to detect subtle morphological changes when comparing follow-up ECGs with the implantation ECG. Purpose To develop and validate a semi-automatic ECG analysis method for evaluating LBBAP ECGs. Methods A semi-automatic ECG analysis method was developed in Python. The software generates a median beat by selecting at least three representative heartbeats from each ECG tracing. R-wave peak time (RWPT) in each lead is computed automatically using Neurokit2. The unipolar pacing spike isidentified as the point of the steepest slope in the ECG signal, and the end of the QRS complex is determined by detecting the return of the waveform to baseline. Results The identified R-peaks, pacing spike, and QRS end point markers are displayed in a graphical user interface (Figure), allowing clinicians to manually adjust the R-peak positions and QRS boundaries if the automated detection is inaccurate. To evaluate the reliability of the ECG analysis method, 48 ECGs from patients undergoing LBBAP for bradycardia (with narrow QRS at baseline and preserved left ventricular ejection fraction) were analyzed by three independent observers, each performing two reviews to assess both inter- and intra-observer agreement. Inter-observer agreement for V6RWPT showed an intraclass correlation coefficient (ICC) of 0.92, and the ICC for the V6-V1 interpeak time was 0.86. Intra-observer ICCs were 0.95 for V6RWPT and 0.99 for the V6-V1 interpeak time. The mean bias between two of the three observers was 0.0 ± 4.9 ms for V6RWPT and 0.3 ± 3.9 ms for the V6-V1 interpeak time. Conclusion The developed semi-automatic ECG analysis method demonstrated good inter- and intra-observer agreement in quantifying LBBAP ECG parameters. By combining automated detection with manual verification, it provides a reliable, reproducible, and efficient tool for standardized assessment of LBBAP ECGs.