Accuracy of left atrial bipolar voltages obtained by ConfiDENSE multielectrode mapping in patients with persistent atrial fibrillation.

Abstract

The ConfiDENSE™ module (Carto3 v4) allows rapid annotation of endocardial electrograms acquired by multielectrode (ME) mapping. However, its accuracy in assessing atrial voltages is unknown. Two ConfiDENSE™ left atrial voltage maps were created during continuous pacing in 20 patients undergoing catheter ablation for persistent AF using a ME lasso catheter and a contact force (CF) sensing ablation catheter. The automated tissue proximity indicator (TPI) filter was then applied to the ME map to yield a TPI map. Reference maps (RM) were created offline by a blinded observer by manually assessing all points against fidelity criteria. Bipolar voltages and proportion of low voltage points (<0.5mV) derived from the ME, CF, and TPI maps were compared with those derived from the RM. Note that 853 ± 365 points, 252 ± 184 points, and 144 ± 73 were collected for ME, TPI, and CF maps, respectively, and 429 ± 153 points were included in the RM. Voltages with CF and TPI maps were similar to those with RM (1.57 ± 0.47mV vs. 1.63 ± 0.31mV, P=0.57 and 1.50 ± 0.38mV vs. 1.63 ± 0.31mV, P=0.07, respectively), whereas ME maps showed a significantly lower mean voltage (1.00 ± 0.22mV, P<0.001). As compared to RM maps (17 ± 8%), low voltage points were significantly overestimated by the ME maps (50 ± 9% (P<0.001) and TPI maps (28 ± 13% (P<0.001), but not by the CF maps (22 ± 14%, P=0.17). Application of the TPI filter to ConfiDENSE maps significantly increases the quality of the voltage data, conserving a reasonable point density, but still overestimates low voltage points as compared to CF-sensing maps or maps reviewed manually.