Session S73.1

The Role of Volume Conductivities in Simulation of Implantable Defibrillators

JG Stinstra*, M Jolley, DH Brooks, JK Triedman, RS MacLeod

University of Utah
Salt Lake City, UT, USA

Commonly raised questions for the implantation of an implantable cardiac defibrillator (ICD) in especially pediatric patients, include which implantable electrode locations offer the best efficacy and what are the effects of variation in the size and conductivity of the torso volume conductor. Electrode placement is constrained by such factors as the size of the devices in comparison to the patient size or a pre-existing lead set that needs to be augmented for a better performance. In order to offer a better understanding of the role of electrode configurations, we are developing a simulation software suite that lets the operator evaluate different electrode configurations in silico. We have applied this system to sets of segmented CT images of the human thorax using the critical mass criterium to predict the efficacy of the defibrillation. The resulting models of the torso contained 10 distinct tissue types, each of which required explicit segmentation that could only be done partially automatic. A more practical clinical approach would be to use of magnetic resonance (MR) images and fewer tissue types. To investigate which tissue types are needed for the simulations, we created a database of three different torso geometries with 49 possible electrode configurations which included both epicardial, transvenous, and subcutaneous electrodes and different locations of the ICD can, which often serves as one of the electrodes. The most elaborate torso model included blood, bone, heart, muscle, fat, kidney, liver, lung, bowel gas, and connective tissue and models of both the electrode wires and the ICD cans that were based on actual shapes of devices and the electrode configurations available. Using this database we repeated a series of simulations in which we varied the selection of tissues that were present in the volume conductor model (1470 simulations in total). For each torso/electrode configuration, we determined the energy required for successful defibrillation and compared this value to that using the fully resolved torso model. For the subcutaneous electrode configurations the most important tissue types were blood, followed by bowel gas, lungs and heart. Omitting blood on average lead to a 20% difference, which was fairly consistent over different electrode configurations (range 29 - 11% difference). A similar pattern was observed for epicardial and transvenous electrode leads, with even a wider range of results when omitting blood. There were also more complex interactions of electrode configurations and torso geometric model, for example bowel gas led to significant differences in energy requirements in configurations that had the ICD can in the lower abdomen (up to 47%). Use of a homogeneous torso resulted in model errors of up to 63%. For errors below 10%, a model that included heart, lung, bone, blood, and bowel gas appeared to be sufficient.

(Abstract Control Number: 143)