Spiral waves characterization: Implications for an automated cardiodynamic tissue characterization

Highlights

• Spiral waves can be clustered using localized electrogram readings obtained with most commonly used multipolar diagnostic catheters.

• Normalized compression distance (NCD) is shown to be a powerful and robust tool in discrimination of distinct properties manifested on a set of EGMs without a need to extract features.

• Compressibility of electrogram dataset is found to be more informative in segregation of spiral wave behaviors than spectral parameter of it.

Abstract

Background and objective: Spiral waves are phenomena observed in cardiac tissue especially during fibrillatory activities. Spiral waves are revealed through in-vivo and in-vitro studies using high density mapping that requires special experimental setup. Also, in-silico spiral wave analysis and classification is performed using membrane potentials from entire tissue. In this study, we report a characterization approach that identifies spiral wave behaviors using intracardiac electrogram (EGM) readings obtained with commonly used multipolar diagnostic catheters that perform localized but high-resolution readings. Specifically, the algorithm is designed to distinguish between stationary, meandering, and break-up rotors. Methods: The clustering and classification algorithms are tested on simulated data produced using a phenomenological 2D model of cardiac propagation. For EGM measurements, unipolar-bipolar EGM readings from various locations on tissue using two catheter types are modeled. The distance measure between spiral behaviors are assessed using normalized compression distance (NCD), an information theoretical distance. NCD is a universal metric in the sense it is solely based on compressibility of dataset and not requiring feature extraction. We also introduce normalized FFT distance (NFFTD) where compressibility is replaced with a FFT parameter.

Results: Overall, outstanding clustering performance was achieved across varying EGM reading configurations. We found that effectiveness in distinguishing was superior in case of NCD than NFFTD. We demonstrated that distinct spiral activity identification on a behaviorally heterogeneous tissue is also possible.

Conclusions: This report demonstrates a theoretical validation of clustering and classification approaches that provide an automated mapping from EGM signals to assessment of spiral wave behaviors and hence offers a potential mapping and analysis framework for cardiac tissue wavefront propagation patterns.

Introduction

Spiral waves are symptoms accompanying cardiac electric rotors observed during fibrillation [1], [2]. Spiral waves or rotors are associated with special types of functional reentrant circuits, a concept considered as one of the sustaining factors for cardiac fibrillatory behavior. In this study, spiral wave behaviors are categorized under three subtypes: stationary, meandering, and breakup. Stable spiral waves (i.e. stationary and meandering) are manifested as a rotational activity around a tip where wavefront and wavetail meets without fragmentation. The tip either pivots in form of negligibly moving or anchoring somewhere (i.e. stationary) or drifts in various trajectories (i.e. meander). Breakup is the fragmentation of spiral waves due to either anatomical or functional interactions. As far as cardiac condition is concerned, stable rotors are mainly associated with localised and more organised electrical activities [3] while rotor breakup corresponds to chaotic and disorganised ones [4].

Mapping studies have shown evidences for the association of spiral waves with cardiac fibrillation [5], [6], [7], [8], [9], [10], [11], [12], [13]. Specifically, both stationary and meandering spiral waves which are associated with cardiac electric rotors have been observed in-vitro [5], in-vivo [9], [10], [11], [12] and rotor break-ups have been revealed through high density mapping of human right atrium [7]. However, special tissue preparations and high density mapping is required for these processes. Also, in-silico studies uses membrane potentials, cellular level data, from the entire tissue [14], [15]. In this study, spiral waves are segregated using localized, smaller number of EGMs which are surface potentials, standard electrophysiological data obtained with most commonly used diagnostic catheters.

In the clinic setting, the most commonly used mapping tools are multipolar diagnostic catheters which provide high-resolution but spatiotemporal recordings with 10–20 electrodes each having a diameter of 15–25 mm. The distance between neighboring electrodes is usually 2 mm. The catheters can be in different shapes including circular as in Lasso (Biosense Webster, Diamond Bar) and Optima (St. Jude Medical, St. Paul) or spiral as in AFocus II (St. Jude Medical, St. Paul) or five-arm (or star) as in PentaRay (Biosense Webster, Diamond Bar). Another less commonly used mapping catheter provides more global but low-resolution recordings with the array of 64 electrodes with 8 splines each constituting of 8 electrodes with 2–3 mm spacing is called basket catheter such as Constellation (Boston Scientific, Natick). However, basket catheters are not widely used due to high cost and added procedural time. While multipolar catheters are widely used, they provide limited mapping information due to localized readings. In this study, a system is proposed that uses localized but high-resolution multipolar catheter readings and provides a spatially extrapolated interpretation of cardiac tissue dynamics in terms of spiral wave behaviors.

Distinct wavefront propagation patterns are manifested through distinct EGMs [16], [17]. Hence, features or quantities extracted from EGM signals [18] and their optimal selection [19] is needed in order to analyze fibrillatory patterns. Considering time varying EGM signals from multi-channels, this task becomes even more difficult. In this paper, we use NCD, a similarity metric, which gives the distance between two objects without a need to extract features or parameters nor does it require background knowledge of the domain of the application [20]. NCD is a state of the art method and based on the computed bit lengths of data files where compression of single, pair, or multisets [21] of data files is performed. It yielded successful results in diverse fields and applications [20]. Details of the technique are covered in Section 2.

In this study, we present an automated grouping as well as classification scheme for spiral wave behaviors through a novel approach that directly uses spatiotemporal EGM readings without feature extraction to assess the distance metrics based solely on compressibility. The workflow including clustering and classification schemes is demonstrated in Fig. 1. Following steps can be summarized: Forward simulations were performed to generate spiral waves that can be classified into three distinct behaviors: stationary, meandering, and break-up. EGM signals were obtained from each simulation as measurement configuration is varied over catheter type used, their location on the tissue, and reading style as being unipolar or bipolar. An information theoretical distance measure between simulated behaviors are assessed using NCD or NFFTD. Clustering evaluations are carried through gap statistics and Jaccard index (JI). Through gap statistics, the number of labels are also automated. Finally, classification analysis within EGM recording configurations and also spiral behavior identification on a behaviorally heterogeneous tissue is carried out. The remainder of this manuscript is organized as follows: In Section 2, methods used for simulation, distance measures, and clustering are presented. In Section 3, results of various tests on the clustering algorithm are presented and discussed. In Section 4, clinical implications of findings with future directions and limitations are discussed. Finally, the paper is concluded with summarization of findings and contributions.