Session SA3.5

An Electrophysiological Cardiac Model with Applications to Ischemia Detection and Infarction Localization

MA Mneimneh*, RJ Povinelli

Marquette University
Milwaukee, WI, USA

A novel electrophysiological cardiac model is introduced in this paper. The proposed cardiac model considers six key regions that characterize the cardiac electrical activity. This allows the model to provide a patient geometry independent solution for the forward and inverse electrocardiology problems in near real time. The major drawback of the current cardiac modeling methods is the computational complexity because they model more than 100, 000 cells of the heart. This complexity does not allow the current techniques to be used in near real time diagnostics. In contrast to the previous models, the proposed cardiac model is used as a basis for two near real time clinical diagnostic applications. The first is the detection of myocardial ischemia. The second is the localization of myocardial infarction. These diagnostic methods use the proposed forward and inverse problem solutions and machine learning approaches to diagnose automatically, noninvasively, and accurately these two serious heart conditions. Moreover, the proposed diagnostic methods have high true positive and negative accuracies suitable to be used in clinical expert systems. The data sets used in this work are the LT-ST database and PTB diagnostics database. The accuracies for the ischemia detection and infarction localization methods are 91% and 68.57%, respectively.

(Abstract Control Number: 109)