Characterisation of electrophysiological conduction in cardiomyocyte co-cultures using co-occurrence analysis

Abstract

Cardiac arrhythmias are disturbances of the electrical conduction pattern in the heart with severe clinical implications. The damage of existing cells or the transplantation of foreign cells may disturb functional conduction pathways and may increase the risk of arrhythmias. Although these conduction disturbances are easily accessible with the human eye, there is no algorithmic method to extract quantitative features that quickly portray the conduction pattern. Here, we show that co-occurrence analysis, a well-established method for feature recognition in texture analysis, provides insightful quantitative information about the uniformity and the homogeneity of an excitation wave. As a first proof-of-principle, we illustrate the potential of co-occurrence analysis by means of conduction patterns of cardiomyocyte–fibroblast co-cultures, generated both in vitro and in silico. To characterise signal propagation in vitro, we perform a conduction analysis of co-cultured murine HL-1 cardiomyocytes and murine 3T3 fibroblasts using microelectrode arrays. To characterise signal propagation in silico, we establish a conduction analysis of co-cultured electrically active, conductive cardiomyocytes and non-conductive fibroblasts using the finite element method. Our results demonstrate that co-occurrence analysis is a powerful tool to create purity–conduction relationships and to quickly quantify conduction patterns in terms of co-occurrence energy and contrast. We anticipate this first preliminary study to be a starting point for more sophisticated analyses of different co-culture systems. In particular, in view of stem cell therapies, we expect co-occurrence analysis to provide valuable quantitative insight into the integration of foreign cells into a functional host system.

Keywords:

• electrophysiology
• microelectrode arrays
• finite element method
• cardiomyocytes
• texture analysis
• pattern recognition

Acknowledgements

We would like to thank Shilpi Roy for her help in establishing cell culture protocols and Dr Jeff Rogers for his intellectual insight into the establishment of the co-occurrence matrix. This work was supported in part by the NSF Graduate Student Research Fellowship (MQC), by the Sang Samuel Wang Stanford Graduate Fellowship (JW), by the NSF CAREER Award CMMI-0952021(EK) and by the California Institute for Regenerative Medicine through cooperative agreement RS1-00232-1 (GTAK).