Efficient semi-automatic segmentation for creating patient specific models for virtual environments
Yifan Song, Ken Brodlie, Andrew J. Bulpitt · White Rose Research Online (University of Leeds, The University of Sheffield, University of York) · 2008
Abstract. There is an increasing demand for the development of virtual environments for training in vascular interventional radiological procedures. This requires fast and precise segmentation of varied abdominal structures from a wide range of image modalities. This paper presents an efficient semi-automatic segmentation system which combines image processing techniques and mathematical morphology operations to obtain an initial segmentation close to the target structure shape. This initial segmentation is then embedded into a level set function to obtain a refined segmentation result. Minimal intervention is required in comparison to other level set based approaches. The approach also dramatically decreases processing time and reduces the risks of leaking at weak boundaries, without compromising the accuracy of the segmentation. 1