Singular value decomposition of optically-mapped cardiac rotors and fibrillatory activity

Our progress of understanding how cellular and structural factors contribute to arrhythmia is hampered in part because of controversies as to whether a fibrillating heart is driven by a single, several, or multiple number of sources, whether they are focal or reentrant and how to localize them. Here we demonstrate how a novel usage of the neutral singular value decomposition (SVD) method enables the extraction of the governing spatial and temporal modes of excitation from a rotor and fibrillatory waves. Those modes highlight patterns and regions of organization in the midst of the otherwise seemingly random propagating excitation waves. We apply the method to experimental models of cardiac fibrillation in rabbit hearts. We show that SVD analysis is able to enhance the classification of the heart electrical patterns into regions harboring drivers in the form of fast reentrant activity and other regions of by-standing activity. This enhancement is accomplished without any prior assumptions regarding the spatial, temporal or spectral properties of those drivers. The analysis corroborates that the dominant mode has the highest activation rate and further reveals a new feature: a transfer of modes from the driving to passive regions resulting in a partial reaction of the passive region to the driving region.