Sparse spectral analysis of atrial fibrillation electrograms

Abstract

Atrial fibrillation (AF) is a common heart disorder. One of the most prominent hypothesis about its initiation and maintenance considers multiple uncoordinated activation foci inside the atrium. However, the implicit assumption behind all the signal processing techniques used for AF, such as dominant frequency and organization analysis, is the existence of a single regular component in the observed signals. In this paper we take into account the existence of multiple foci, performing a spectral analysis to detect their number and frequencies. In order to obtain a cleaner signal on which the spectral analysis can be performed, we introduce sparsity-aware learning techniques to infer the spike trains corresponding to the activations. The good performance of the proposed algorithm is demonstrated both on synthetic and real data.

Blind analysis of atrial fibrillation electrograms: A sparsity-aware formulation

“…2 Note that Eq. (16) corresponds to an energy-normalized Gaussian instead of the standard unnormalized Gaussian used in [15,20]. The derivation of the normalized Gaussian, as well as the mexican hat wavelet of Eq.…”

“…Hence, since the spectral analysis is applied to the inferred sparse activation sequence, this approach is called sparse spectral analysis (SSA) [20]. The number of peaks extracted (after the post-processing described in Section 5.2) is an estimate of the number of existing foci, and their locations an estimate of their frequencies.…”

“…2. Inferring the spike trains associated to the activation times using a LASSO [33], followed by a post-processing step to ensure that biological restrictions are met [20]. 3.…”

“…Applying a late post-processing step in order to eliminate harmonics and subharmonics. A preliminary version of this paper has appeared in [20]. When compared to [20], this contribution includes several theoretical and practical modifications: the pre-processing stage has been enhanced, the choice of the reconstruction dictionary is briefly discussed and a novel wavelet-based dictionary developed, a method for estimating the noise variance has been incorporated and the spectral analysis stage has also been improved.…”

“…A preliminary version of this paper has appeared in [20]. When compared to [20], this contribution includes several theoretical and practical modifications: the pre-processing stage has been enhanced, the choice of the reconstruction dictionary is briefly discussed and a novel wavelet-based dictionary developed, a method for estimating the noise variance has been incorporated and the spectral analysis stage has also been improved. However, the most important extensions in the paper are in the application side: more realistic simulations on synthetic data have been performed using a typical activation shape for EGMs [8], and applications on real data have been performed for several patients under sinus rhythm and AF conditions.…”