A hybrid surface-volume integral-equation method for analyzing scattering from voxel-based anatomical human models with smooth skin

Abstract

An increasing number of anatomically accurate voxel-based human models, developed from medical images, are becoming available for bioelectromagnetic analysis (J. W. Massey et al, Proc. 34 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">th</sup> Annu. Mtg. Bioelectromagn. Soc, June 2012). Because the stair-cased boundaries in these models limit the accuracy of the numerical analysis, smooth surface-based human models are desired. Although it is relatively simple to extract one organ/tissue from a voxel model and obtain a smoothed surface model by applying a surface construction algorithm (W. Lorensen and H. Cline, Proc. SIGGRAPH, July 1987), the resulting independently processed surfaces are in general not compatible (e.g., they can intersect) and therefore cannot be easily combined to obtain a surface-based human model. Case studies with multilayer spherical head phantoms show that a significant portion of the error in voxel-based models occurs in the outermost layer. Thus, accuracy improvements can be achieved even if only the skin of the human model is smoothed. This, however, gives rise to a different compatibility issue - matching the unstructured, triangulated mesh of the skin to the underlying voxel mesh of the body. In this article, an integral-equation approach is used to circumvent this limitation and benefit from voxel-based models with improved skin.