Analysis of spectrogram parameter organization applied to the characterization of atrial fibrillation

Atrial Fibrillation (AF) is the most common arrhythmia encountered at advanced age. The possibility of applying signal analysis to the electrocardiogram (ECG) in order to distinguish between terminating and non-terminating episodes of AF would be very helpful in the regular clinical practice. In this work, an analysis of spectrogram parameters organization of the Atrial Activity (AA) is carried out in order to classify between terminating and non-terminating AF episodes. The main peak frequencies, second largest peak frequencies and their respective peak magnitudes are extracted using cubic spline fitting and several numerical series are constructed from them. The independent samples mean comparison of their Sample Entropy shows five of the constructed sequences to be relevant in the characterization of AF. The bilateral significance obtained by the Studentpsilas t test is less than 0.005 in five of the eight analyzed parameters.