Deeplearningassisted detection and termination of spiral and brokenspiral waves in mathematical models for cardiac tissue
Abstract
Unbroken and brokenspiral waves, in partialdifferentialequation (PDE) models for cardiac tissue, are the mathematical analogs of lifethreatening cardiac arrhythmias, namely, ventricular tachycardia and ventricularfibrillation. We develop (a) a deeplearning method for the detection of unbroken and brokenspiral waves and (b) the elimination of such waves, e.g., by the application of lowamplitude control currents in the cardiactissue context. Our method is based on a convolutional neural network (CNN) that we train to distinguish between patterns with spiralwaves S and without spiralwaves NS . We obtain these patterns by carrying out extensive direct numerical simulations of PDE models for cardiac tissue in which the transmembrane potential V , when portrayed via pseudocolor plots, displays patterns of electrical activation of types S and NS . We then utilize our trained CNN to obtain, for a given pseudocolor image of V , a heatmap that has high intensity in the regions where this image shows the cores of spiral waves and the associated wavefronts. Given this heatmap, we show how to apply lowamplitude currents of a twodimensional Gaussian profile to eliminate spiralwaves efficiently. Our in silico results are of direct relevance to the detection and elimination of these arrhythmias because our elimination of unbroken or brokenspiral waves is the mathematical analog of lowamplitude defibrillation.
 Publication:

Physical Review Research
 Pub Date:
 May 2020
 DOI:
 10.1103/PhysRevResearch.2.023155
 arXiv:
 arXiv:1905.06547
 Bibcode:
 2020PhRvR...2b3155M
 Keywords:

 Physics  Biological Physics;
 Nonlinear Sciences  Pattern Formation and Solitons
 EPrint:
 Phys. Rev. Research 2, 023155 (2020)