Clinically, bipolar electrograms (EGMs) are interpreted based on time and voltage characteristics such as amplitude and fractionation index. However, this discards much of the information within the signal. EGMs can also be analysed in the frequency and time-frequency domain. Omnipolar EGMs are a direction-independent form of bipolar EGMs and can be calculated from unipolar EGMs. Using an ex-vivo porcine preparation in combination with unipolar and omnipolar EGMs, we sought to determine whether frequency and time-frequency features can be used to differentiate between substrate at baseline and with repolarisation abnormalities.
Unipolar EGMs were recorded from hearts paced at varying cycle lengths between 300 and 1,500 ms at positions across the left ventricle and left atrium. EGMs were measured at baseline (n=7) and post-modulation with KATP channel agonist, pinacidil (n=2). Omnipolar EGMs were calculated from the unipolar EGMs using a custom-written algorithm. A total of 8 frequency domain features (dominant frequency [DF], power spectral density [PSD] max, PSD mean, PSD median, PSD standard deviation [SD], PSD kurtosis, PSD skew and root mean square [RMS] amplitude) and 8 time-frequency domain features (max modulus, variance of signal energy, scale corresponding to minimum and maximum energy, pseudo frequency corresponding to minimum and maximum energy, and percentage of energy corresponding to max modulus) were extracted using a fast Fourier transform and continuous wavelet analysis of the EGM signals, respectively.
In the ventricle, both unipolar and omnipolar EGMs showed significant difference between baseline vs pinacidil values for all frequency and time-frequency domain features. For frequency domain features (Figure 1a), unipolar EGMs recorded a greater mean negative percentage change from baseline with pinacidil for four4 features (PSD SD, RMS amplitude, PSD kurtosis and PSD skew) while omnipolar EGMs recorded a greater mean positive percentage change from baseline with pinacidil for the remaining 4 features (DF, PSD mean, PSD median, PSD max). For time-frequency domain features (Figure 1b), unipolar EGMs recorded a greater negative for 2 features (variance of signal energy and max modulus) while omnipolar EGMs recorded the same for percentage of energy corresponding to max modulus.
Frequency and time-frequency domain features can identify KATP channel opening in both unipolar and omnipolar EGMs. With further study on different pro-arrhythmic modulators, such as gap junction or sodium channel blockers, this approach may provide mechanistic links between the EGM and properties of the myocardium and may help guide future ablation procedures. Omnipolar EGMs are not found to be superior to unipolar EGMs in this context.