Time-frequency characterization of atrial arrhythmias using principal decomposition

A new noninvasive technique for atrial arrhythmia analysis is presented which, based on time-frequency analysis and principal decomposition, produces trends of the atrial signal characteristics using the surface ECG. New homogeneity measures are introduced in order to continuously measure how well the decomposed functions represent the atrial signals. A database containing signals from 20 patients with different atrial arrhythmias (mainly atrial fibrillation) was analyzed. It was found that the method is very well-suited for characterizing the signals and it is expected that the resulting functions may be useful for separating different arrhythmias.