ABSTRACT
A proof of concept workflow is demonstrated to easily translate 3D medical image data into finite element (FE) simulation ready phantom models. First, novel methods areused to drastically reduce manual segmentation time for a virtual population. Next, using Simpleware software, the segmented voxel datasets are extracted into faceted 3D CAD objects for tissues, while simultaneously maintaining conformal multi-tissue interfaces. Finally, the 3D CAD geometries are demonstrated to be readily compatible in a commercial 3D electromagnetic simulator, ANSYS HFSS.
NOMENCLATURE
Simulation, Meshing, Human Phantoms, Segmentation
INTRODUCTION
3D image-based meshing of multi-part structures from medical scan data (PET, SPECT, CT, MRI) continues to open up exciting new possibilities for the application of electromagnetic (EM), finite element (FE), and computational fluid dynamics (CFD) methods to a wide range of biomedical problems [1]. However, significant challenges to creating a population of simulation compatible human models have prevented them from becoming readily available for industry. These include: 1) Dataset Availability – Due to health care industry privacy rules and the cost of creating a virtual population of MRI or CT scan data, very few readily available dataset repositories of human phantoms exist for industry use [2,3,4]. 2) Segmentation Difficulty – Segmentation of scan datasets is extremely man-hour intensive. Depending on images and/or segmentation quality, effort is often measured by man-months or years for a single model. This time constraint has precluded patient specific and/or virtual population FE modeling. 3) Clean CAD Model Extraction – If the CAD model contains coincident spline (NURBS) surface interfaces between tissues, meshing may be slowed significantly or prevented completely. Likewise, faceted volumetric meshes and CAD geometry must contain conformal face mapping between touching objects. Since traditional part-by-part meshing approaches risk gaps between, or node penetration into adjacent parts, manual and time consuming repair is required.
This paper demonstrates a potential solution to these challenges through a fast and efficient workflow that begins with newly available anatomical geometries, and culminates in a solved multi-object computational simulation. Using the new series of 4D extended cardiac-torso (XCAT) phantoms created by Segars and colleagues [5], we use ScanIP (Simpleware Ltd.,Exeter, UK) to convert these datasets into multi-object simulation ready geometry files that are imported into HFSS(ANSYS Inc., Canonsburg, PA) for EM simulation and analysis.
``````
沪公网安备 31011702000137号 
营业执照
微信
新浪微博