Significance of mixing matrix structure on principal component-based analysis of atrial fibrillation body surface potential maps

Atrial fibrillation (AF) type classifiers are still hardly accepted in clinical practice due to their invasive approach. In this work a new automated method to assess noninvasively different AF types is presented, based on the high spatial resolution given by body surface potential maps (BSPM). AF organization degree was assessed by its influence on the spatio-temporal pseudo-stationarity and complexity of a principal component analysis mixing matrix repeatedly derived along a BSPM recording. Stationarity was analyzed in terms of ability of the mixing matrix derived for a specific recording segment to retrieve the AA components of subsequent segments, while complexity in terms of its number of significant components. Results show that AF organization is reflected in a greater pseudo-stationarity of the mixing matrix along the recordings and in a lower number of components needed to represent the AA, interpreted as a lower complexity in the underlying AA in patients with organized AF type I.