Darrell Swenson

Subject-specific, multiscale simulation of electrophysiology: a software pipeline for image-based models and application examples

Many simulation studies in biomedicine are based on a similar sequence of processing steps, starting from images and running through geometric model generation, assignment of tissue properties, numerical simulation and visualization of the results—a process known as image-based geometric modelling and simulation. We present an overview of software systems for implementing such a sequence both within highly integrated problem-solving environments and in the form of loosely integrated pipelines. Loose integration in this case indicates that individual programs function largely independently but communicate through files of a common format and support simple scripting, so as to automate multiple executions wherever possible. We then describe three specific applications of such pipelines to translational biomedical research in electrophysiology.

A toolkit for forward/inverse problems in electrocardiography within the SCIRun problem solving environment

Computational modeling in electrocardiography often requires the examination of cardiac forward and inverse problems in order to non-invasively analyze physiological events that are otherwise inaccessible or unethical to explore. The study of these models can be performed in the open-source SCIRun problem solving environment developed at the Center for Integrative Biomedical Computing (CIBC). A new toolkit within SCIRun provides researchers with essential frameworks for constructing and manipulating electrocardiographic forward and inverse models in a highly efficient and interactive way. The toolkit contains sample networks, tutorials and documentation which direct users through SCIRun-specific approaches in the assembly and execution of these specific problems.

Spatial organization of acute myocardial ischemia

INTRODUCTION Myocardial ischemia is a pathological condition initiated by supply and demand imbalance of the blood to the heart. Previous studies suggest that ischemia originates in the subendocardium, i.e., that nontransmural ischemia is limited to the subendocardium. By contrast, we hypothesized that acute myocardial ischemia is not limited to the subendocardium and sought to document its spatial distribution in an animal preparation. The goal of these experiments was to investigate the spatial organization of ischemia and its relationship to the resulting shifts in ST segment potentials during short episodes of acute ischemia. METHODS We conducted acute ischemia studies in open-chest canines (N=19) and swines (N=10), which entailed creating carefully controlled ischemia using demand, supply or complete occlusion ischemia protocols and recording intramyocardial and epicardial potentials. Elevation of the potentials at 40% of the ST segment between the J-point and the peak of the T-wave (ST40%) provided the metric for local ischemia. The threshold for ischemic ST segment elevations was defined as two standard deviations away from the baseline values. RESULTS The relative frequency of occurrence of acute ischemia was higher in the subendocardium (78% for canines and 94% for swines) and the mid-wall (87% for canines and 97% for swines) in comparison with the subepicardium (30% for canines and 22% for swines). In addition, acute ischemia was seen arising throughout the myocardium (distributed pattern) in 87% of the canine and 94% of the swine episodes. Alternately, acute ischemia was seen originating only in the subendocardium (subendocardial pattern) in 13% of the canine episodes and 6% of the swine episodes (p<0.05). CONCLUSIONS Our findings suggest that the spatial distribution of acute ischemia is a complex phenomenon arising throughout the myocardial wall and is not limited to the subendocardium.

Evaluating the effects of border zone approximations with subject specific ischemia models

Current computational models of acute ischemia are deficient because of their inability to be validated against experimental data and their lack of geometric realism. Past models of ischemia have been based on geometric primitives or hearts for which no electrical measurements exist. One consequence is that it is necessary to make modeling assumptions that are not supported by measurements or direct validation with experiments. Based on our subject specific simulations and measurements, we hypothesize that assumptions about the nature and scale of the border zone play a significant role in determining the cardiac potential distributions from ischemic sources of injury current. Geometrically accurate models were created from Magnetic Resonance Imaging (MRI) and Diffuser Tensor Imaging (DTI) after an in situ canine ischemia experiment. The ischemic zone was defined based on transmural electrode readings and was used in a static bidomain simulation representing a time point during the ST segment. Varying the width of the border zone in the simulations changed the magnitude and distribution of epicardial depressions and elevations. We also found that a border zone with linear variation of potential from healthy to ischemic regions was not adequate to simulate measured potentials. A more sophisticated border zone that included an explicit region of partial ischemia was necessary to simulate the field gradients seen in experimental data.

Sensitivity of epicardial electrical markers to acute ischemia detection

INTRODUCTION We hypothesize that electrocardiographic measurements from the intramyocardial space contain more sensitive markers of ischemia than those detectable on the epicardium. The goal of this study was to evaluate different electrical markers for their potential to detect the earliest phases of acute myocardial ischemia. METHODS We conducted acute ischemia studies in open chest animal, by creating finely controlled demand or supply ischemic episodes and recording intramyocardial and epicardial potentials. RESULTS Under the conditions of mild perfusion deficit, acute ischemia induced changes in the T wave that were larger and could be detected earlier on the epicardial surface than ST-segment changes. CONCLUSIONS Our findings indicate that in the setting of very acute ischemia, epicardial T waves have higher sensitivity to mild degrees of acute ischemia than epicardial ST potentials. These results suggest that changes in the T wave shape may augment shifts in ST segments to improve ECG based localization of ischemia.

Sensitivity and specificity of substrate mapping: An in silico framework for the evaluation of electroanatomical substrate mapping strategies

Voltage mapping is an important tool for characterizing proarrhythmic electrophysiological substrate, yet it is subject to geometric factors that influence bipolar amplitudes and thus compromise performance. The aim of this study was to characterize the impact of catheter orientation on the ability of bipolar amplitudes to accurately discriminate between healthy and diseased tissues.

Impacts of Boundary Conforming Meshes on Electrical Cardiac Simulation

Computational simulation has become an indispensable tool in the study of both basic mechanisms and pathophysiology of all forms of cardiac electrical activity. Such simulations depend heavily on geometric models that are either realistic or even patient specific. These models consist of a connected mesh of sometimes millions of polygonal elements that must capture the complex external shapes and internal boundaries among regions of the heart. The resulting meshes can be non-conforming, i.e., they have element faces that fail to align with the tangents of the surfaces or boundaries and consequently the elements are a poor approximation of these smooth surfaces and boundaries. We hypothesize that such jagged, non-conforming meshes, which are often preferred, as they are easier to create, produce local artifactual concentrations of current that lead to simulation errors large enough to distort the resulting potential fields and generate misleading results. We tested this hypothesis on two types of numerical approximation used in bioelectric simulations: bidomain, and reaction-diffusion bidomain. Comparison with gold standard results for the monodomain and bidomain simulations showed that errors within a few elements (3-5) of the surface could be as large as 10-32%. The root mean squared error over the entire mesh was more modest, ranging from 1-6%. In the case of reaction diffusion simulations, by contrast, such meshing errors accounted for only an insignificant component of overall simulation uncertainty. These findings lead to the conclusion that while non-conforming meshes are certainly less costly to produce, their use can result in substantial local errors that depend highly on the specific problem of interest and the numerical approximation approach.