Spiral waves characterization: Implications for an automated cardiodynamic tissue characterization. (July 2018)

• Record Type: Journal Article
• Authors: Alagoz, Celal; Cohen, Andrew R.; Frisch, Daniel R.; Tunç, Birkan; Phatharodom, Saran; Guez, Allon
• Is Part Of: Computer methods and programs in biomedicine. Volume 161(2018)
• Page Start: 15
• Page End: 24
• Publication Date: 2018-07
• Subjects: Spiral waves -- Clustering -- Classification -- Cardiac fibrillation -- Cardiac electro-physiology simulation -- Intracardiac electrograms -- Normalized compression distance; Medicine -- Computer programs -- Periodicals; Biology -- Computer programs -- Periodicals; Computers -- Periodicals; Medicine -- Periodicals; Médecine -- Logiciels -- Périodiques; Biologie -- Logiciels -- Périodiques; Biology -- Computer programs; Medicine -- Computer programs; Periodicals; Electronic journals; 610.28
• DOI: 10.1016/j.cmpb.2018.04.006
• Languages: English
• ISSNs: 0169-2607

Abstract

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.