Differentiating endocardial from epicardial signal using spectral patterns of bipolar ventricular electrograms

Background High-frequency components in intracardiac electrograms may reflect tissue origin and conduction properties. We assessed spectral features in ventricular endocardial electrograms from the endocardial and epicardial activation origins. Methods We used 8 open-chest swine models. Electrical stimulation was performed at endocardial (Endo) and epicardial (Epi) areas of the left ventricle in each model using a needle electrode. Endocardial electroanatomical maps were created using a DecaNav catheter at each stimulation session. Bipolar electrograms were recorded within 10 mm distant sites from the earliest activation point of each electrical stimulation session with a sampling rate of 1000 Hz and bandpass filter of 30-500 Hz, and analyzed using Welch’s power spectral density. Frequency bands of 0–100, 100–200, 200–300, 300–400, and 400–500 Hz were evaluated. Efficacy of Endo/Epi discrimination was validated using additional 3 animals. Results A total of 449 electrograms were analyzed. Epicardial signals showed higher power in the 0–100 Hz band (93.7% vs. 92.2%, P < 0.01), while exhibiting lower power in the 100–500 Hz bands. The highest discriminatory AUC (0.644, cut-off value 0.271%, sensitivity 59.0%, specificity 67.7%) was observed in the 300–400 Hz band (Figure). In validation study, electrograms of 15 sites were tested under the criterion of power ratio of 300-400 Hz > 0.271% for Endo sites. The sensitivity and specificity of that criterion were 71% and 38%, respectively. Conclusion Spectral analysis of electrograms shows moderate utility in distinguishing endocardial from epicardial origins. Power in the 300–400 Hz range appears most effective, suggesting possible utility in clinical localization.