You are here

Efficient Model Posing and Morphing Software


OBJECTIVE: Develop an efficient software approach to manipulate the pose and morph the shape of voxelized anatomical models. These models are high-fidelity (mm resolution) representations of the tissue and organ layout within the body. DESCRIPTION: Electromagnetic devices are used increasingly in society, with applications in communication, medicine, security, and defense, among other disciplines and technology areas. This has led to a great deal of research regarding the safety and potential health hazards of such devices. To aid in this study, several high-fidelity voxelized anatomical body models have been created. These voxel models can be used in conjunction with sophisticated computational electromagnetic (EM) and thermal solvers to address the energy absorption rates and temperature increases expected within tissue due to radio frequency (RF) exposures. Recently, researchers have created extremely efficient EM and thermal solvers through hardware acceleration or approximation techniques. These software approaches allow researchers to quickly simulate (within minutes) the energy absorption and temperature increases within tissue for a single anatomical model. However, simulation results reported in the scientific literature indicate that the posture of an individual within an incident RF field has a significant effect on model predictions, as does the anatomical geometry. Since voxel models are costly to create, and available postures are limited by the imaging modality used in their generation (typically MRI), research has focused on posture manipulating and morphing software to address this limitation. However, the reposing and morphing software created to date requires up to many hours of computational runtime to achieve a single new voxel model, clearly forming a bottleneck for RF dosimetry studies in comparison to the ultrafast EM and thermal solvers. Therefore, there is a need for the development of highly efficient software to 1) manipulate the pose of voxelized anatomical models and 2) morph anatomical models according to various anthropometric parameters. The ideal solution should require on the order of minutes or less for the generation of a newly posed or morphed anatomical model. Approaches to the posture manipulation component of this work should allow for realistic manipulation of joints, and seek to conserve mass and dielectric continuity for both bones and associated soft tissues in the vicinity of joints. Additionally, incorporating human motion-capture data would be highly desirable in order to both automate model postures (walking, sitting, etc.) and to ensure high biofidelity. For example, data from the biomechanics research field in general can be utilized to ensure model postures accurately reflect real-life human body postures. Specifically, 3D Anthropometric databases may be useful for obtaining the external dimensions of humans in various poses. Approaches to the model morphing portion of this work should allow models to be morphed according to certain anthropometric parameters (e.g., height, body fat percentage, BMI). Additionally, techniques to allow higher-fidelity partitioning of organs of interest (e.g., re-meshing the brain to differentiate regions) are desirable. Biomedical scientists, health and medical physicists, and bioenvironmental engineers would all benefit from software that enabled efficient voxel model reposing and morphing. Additionally, the techniques developed could be very useful within a surgery planning tool. PHASE I: Determine the computational methods to be used, and develop prototype software that illustrates the effectiveness of the algorithms chosen for both reposing and morphing existing anatomical models. The software may manipulate the models directly as voxel descriptions, or through more advanced boundary representation formats. However, the software output should be converted to a voxel format. PHASE II: Extend the software created in Phase I to be fully developed and optimized with respect to computational runtime. A user interface should be included for manipulating the anatomical models and viewing the output. Where appropriate, metrics should be determined and utilized to verify the software output (e.g., conservation of mass). Preferably, the software should be able to manipulate the anatomy and output voxel models at 1 mm or finer resolution. PHASE III: Use by engineers and health physicists to study risks of accidental RF overexposures over a broad set of exposure conditions. Used by Air Force to predict potential of overexposure during engagement of novel directed energy systems. REFERENCES: 1. V Singh, and D Silver. Interactive Volume Manipulation with Selective Rendering for Improved Visualization. IEEE Symposium on Volume Visualization and Graphics, 2004. 2. T Uusitupa, I Laakso, S Ilvonen, and K Nikoskinen. SAR variation study from 300 to 5000 MHz for 15 voxel models including different postures. Physics in Medicine and Biology, 2010, Vol. 55. 3. P Crespo-Valero, M Christopoulou, M Zefferer, A Christ, P Achermann, K Kikita, and N Kuster. Novel Methodology to characterize electromagnetic exposure of the brain. Physics in Medicine and Biology, 2011, Vol. 56. 4. Z Cheng, S Mosher, J Parakkat, and K Robinette. Human Modeling and Simulation with High Biofidelity. Proceedings of the 2nd International Conference on Applied Digital Human Modeling, July 2012. 5. Robinette, K., Blackwell, S., Daanen, H., Fleming, S., Boehmer, M., Brill, T., Hoeferlin, D., and Burnsides, D., Civilian American and European Surface Anthropometry Resource (CAESAR), Final Report, Volume I: Summary, Technical Report AFRL-HE-WP-TR-2002-0169, National Technical Information Service Accession No. ADA406704, United States Air Force Research Laboratory, 2002.
US Flag An Official Website of the United States Government