Information Fusion for Real-time Motion Estimation in Image-guided Breast Biopsy Navigation

Abstract

A real-time delivery of accurate soft-tissue intervention navigation information is one of the most crucial aspects for accepting the soft-tissue navigation systems for intra-operative use. Currently, soft-tissue navigation systems face some obstacles in terms of registration of the virtual navigation information on the deformable soft-tissue organ. Most of them perform a rigid registration between the virtual data and the organ, and then provide the surgeon with all navigation information. However, they suffer from the disadvantage that the virtual information is not correctly registered to the deformable organ. In order to enable a real-time non-linear registration between the virtual navigation information and the deformable organ, we incorporate different means for tracking the soft- tissue internal and on-surface local motion. Furthermore, we introduce an intelligent information fusion engine for combining the various soft-tissue local motion tracking information into a global motion information channel. The fusion engine is the interface to the motion measurements, a motion dynamics model, and static shape infor- mation, which are combined to compute the a posteriori estimate of the current state of the deformed shape. The dynamics model is realized as a finite-element deformation simulation. In order to test the feasibility of our devised information fusion engine, we have employed it for capturing the global motion of a breast phantom during an image-guided biopsy. The biopsy planning navigation data, in the form of a prior diagnostic MRI, is continuously updated over time according to the a posteriori estimate of the global motion. As a result, the real-time changes in the shape of the breast are always reflected in the biopsy navigation information.