Reality-based Modeling of Deformable Soft Objects via Internal Sensing

点击翻译按钮获取中文摘要

The goal of this research is to establish reality-based modeling, which build physical models of deformable soft objects such as biological tissues and food materials via internal sensing of the objects. Since such deformable soft objects often exhibit non-uniform deformation properties, we establish a modeling technique based on the three-dimensional measurements of inner deformation of a deformable soft object. First, we establish a method to estimate the inner deformation field from the measured MR images of a deformable soft object. We proposed a registration method between two MR volume images before and after the deformation as well as a method to search corresponding points between the two volume images via bidirectional matching. These methods were experimentally evaluated. Additionally, we experimentally evaluated a micro force/moment sensor embedded in a deformable soft fingertip. Second, we established a method to identify non-uniform elastic property of a deformable soft object via internal sensing. We proposed to push a target soft object by another object of which property is known in advance to identify the property of the target object by measuring the inner deformation of the two objects. We proposed a recursive identification method and a batch identification method to evaluate the two methods through simulation. The two methods were experimentally evaluated by using the inner deformation of a phantom measured by a CT scanner. The methods were applied to MR images as well. Next, we developed adaptive techniques for modeling of deformable. soft objects. An adaptive mesh method, where mesh size is changed adaptively according to a mechanical quantity, was applied to cutting operation. A method to calibrate a hierarchical modeling using force/shape sequence was proposed. The proposed models were applied to active ischemia dynamics and mammography.