Spatial complexity and spectral distribution variability of atrial activity in surface ECG recordings of atrial fibrillation.

Considerable research effort has been devoted to the estimation of the degree of organisation of atrial fibrillation (AF), to potentially support clinical decision making. The aims of this study were to: (1) analyse the temporal variability of spatial organisation (complexity) and spectral distribution of AF in body surface potential maps (BSPM), proposing an automated implementation of the analysis and (2) assess the applicability to reduced lead-sets. Twenty-one persistent AF recordings of 3 min each (64 BSPM: 32 anterior, 32 posterior) were analysed. The relationship between spatial organisation (C) and its variability (CV) was quantified on automatically delineated TQ segments. The relationship between spectral concentration (SC) and spectral variability (SV) was quantified on the atrial activity (AA) extracted using principal component analysis. Three different lead-sets: 64, 32 anterior and 10 anterior channels were considered. Significant (p < 0.001) correlation (ρ) was found: ρ(CV, C) ≥ 0.80, ρ(SC, SV) ≤-0.83 for all lead-sets. The results suggest that a higher degree of spatial organisation is associated with reduced variability of spatial organisation over time, and lower spectral variability associated with more prominent spectral peak in the AF frequency band (4-10 Hz).