Dominant Frequency and Organization Index for Substrate Identification of Persistent Atrial Fibrillation

The combined use of dominant frequency (DF) and organization index (OI) might help to identify atrial regions with organized, fast activation rates in persistent atrial fibrillation (persAF). We determined adaptive thresholds for DF and OI based on electrophysiological responses following AF substrate modification. 2048-channel electrograms (3206 EGMs, 30 s, EnSite Array) were analyzed from 10 persAF patients undergoing DF-guided ablation. After QRST subtraction, fast Fourier transform was used to calculate DF and OI. AF cycle length (AFCL) was measured before and after each ablation point (left atrium appendage). EGMs were grouped in two classes: collected at regions whose ablation resulted in AFCL increase $(\geq 10\ ms)$ and AFCL non-increase $( < 10\ ms)$. Patient-specific z-score DF (DFz) and IO(OIz) were tested to separate the two classes (individually and AND-logic). Informedness (J), accuracy (Acc) and F1 score were used to assess classification performance. Best individual classifications were $DFz=0.52 (J=0.16, Acc=65\%, F1=0.41)$, and $OIz=0.60 (J=0.19, Acc=70\%,F1=0.40)$. Best AND-logic (DFz and OIz) classification was $DFz=-0.52$ and $OIz=0.49(J=0.23,Acc=71\%,F1=0.43)$. DF and OI combination might help in the identification of patient-specific AF substrate to guide ablation in future clinical studies.