Reliable path for virtual endoscopy: ensuring complete examinationof human organs

摘要

Virtual endoscopy is a computerized, noninvasive procedure forndetecting anomalies inside human organs. Several preliminary studiesnhave demonstrated the benefits and effectiveness of this modality.nUnfortunately, previous work cannot guarantee that an existing anomalynwill be detected, especially for complex organs with multiple branches.nIn this paper, we introduce the concept of reliable navigation, whichnensures the interior organ surface is fully examined by the physiciannperforming the virtual endoscopy procedure. To achieve this, we proposencomputing a reliable fly-through path that ensures no blind areas duringnthe navigation. Theoretically, we discuss the criteria of evaluating anreliable path and prove that the problem of generating an optimalnreliable path for virtual endoscopy is NP-complete. In practice, wendevelop an efficient method for the calculation of an effective reliablenpath. First, a small set of center observation points are automaticallynlocated inside the hollow organ. For each observation point, therenexists at least one patch of interior surface visible to it, but thatncannot be seen from any of the other observation points. These chosennpoints are then linked with a path that stays in the center of thenorgan. Finally, new points inside the organ are recursively selected andnconnected into the path until the entire organ surface is visible fromnthe path. We present encouraging results from experiments on severalndata sets. For a medium-size volumetric model with several hundrednthousand inner voxels, an effective reliable path can be generated innseveral minutes