Localizing Rotors in Human Atrial Fibrillation Using Differential Entropy.

Rotors sustain atrial fibrillation (AF) and are targeted during AF ablation with variable results. Although Shannon entropy was shown to identify rotors of spiral waves in experimental settings, it fails to detect rotors in human atrial fibrillation (AF). We previously showed that the accuracy of identification sensitively depends on the number of rotors and spatial resolution of signal recording. Another possible limitation of the approach is discretization of probability distribution from continuous unipolar electrograms. Therefore, we assessed the value of differential entropy (DE), which does not require discretization of probability distribution, to locate rotors of human AF using intracardiac 64-electrode basket catheter recordings from 33 patients. The area of maximum DE and rotors precisely overlapped in 9% of the recordings, and was one electrode away in 36%. When the maximum DE was defined as the highest 5^th percentile, the area of maximum DE and rotors precisely overlapped in 22% of the recordings, and was one electrode away in 59%. We conclude that DE is a promising metric to locate rotors in human AF, but the accuracy is modest.