A Pipeline for Generating Physiological Volumetric Ecg Signals

• Published in: Journal of electrocardiology Vol. 49; no. 6; p. 938
• Main Authors: Tate, Jess, Kindall, Brianna, Gillette, Karli, Burton, Brett, Coll-Font, Jaume, Erem, Burak, White, Dan, Khan, Ayla, van Dam, Peter, Wilkinson, Jeff, Simha, Narendra, MacLeod, Rob
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 01.11.2016; Elsevier Science Ltd
• Subjects: Cardiovascular; Cardiovascular disease; Electrocardiography; Simulation
• ISSN: 0022-0736; 1532-8430
• DOI: 10.1016/j.jelectrocard.2016.09.042

Simulation of electrocardiographic signals from normal and pathophysiological hearts presents a new opportunity to meet a growing demand in research and device development. The demand arises from clinically motivated research and commercial device development to improve detection and localization of, for example, regions of ischemia, abnormal conduction, or arrhythmic substrate. The challenge arises due to limited access, especially to measured intracardiac and torso volume potentials in both humans and animal models. Accurate simulation can fill in these missing gaps, leveraging the measured data that are actually available as well as information from non-invasive imaging techniques to create highly realistic and flexible models. We have developed a complete pipeline of algorithms and software tools to generate electrocardiographic data in human torso geometries for use in a variety of applications. Our pipeline consists of three main steps: compilation and generation of cardiac source data, registration of the cardiac data to various torso geometries, and predicting the resulting torso potentials. The first step involves compiling a database of cardiac source electrograms, which can be tailored to the scope of inquiry. The next step involves registering each source geometry to each torso geometry. The final step of generating the electrocardiographic signals through the torso provides electrocardiograms from anywhere within the torso. Each step is modular, which maximizes flexibility at each stage. Our pipeline brings together various data types, software, and algorithms. The database is compiled from recordings from animal experiments or from biophysical simulations using software such as CARP or ECGSim. The registration step merges geometries to a common framework using a modified iterative closest point method. The final step of signal generation uses finite and boundary element approaches from the Forward/Inverse Toolkit in SCIRun. All these elements are combined to maintain flexibility, yet optimize efficiency of the pipeline. Using our pipeline, we can combine experimental animal recordings and simulated cardiac potentials with human geometries to create clinically relevant data in any location of the torso. By using this pipeline to simulate electrocardiographic fields from ischemia, PVCs, fibrillation, and other arrhythmias, we can provide the data needed to improve detection and location, and thus treatment for many cardiac disorders.