Atrial fibrillation subtypes classification using the General Fourier-family Transform

Abstract Atrial fibrillation patients can be classified into paroxysmal, persistent and permanent attending to the temporal pattern of this arrhythmia. The surface electrocardiogram hides this differentiation. A classification method to discriminate between the different subtypes of atrial fibrillation by using short segments of electrocardiograms recordings is presented. We will process the electrocardiograms (ECGs) using time–frequency techniques with a global accuracy of 80%. Real cases are evaluated showing promising results for an implementation in a semiautomated diagnostic system.