Avoiding Mesh Folding in 3D Optimal Surface Segmentation
The segmentation of 3D medical images is a challenging problem that benefits from incorporation of prior shape information. Optimal Surface Segmentation (OSS) has been introduced as a powerful and flexible framework that allows segmenting the surface of an object based on a rough initial prior with robustness against local minima. When applied to general 3D meshes, conventional search profiles constructed for the OSS may overlap resulting in defective segmentation results due to mesh folding. To avoid this problem, we propose to use the Gradient Vector Flow field to guide the construction of non-overlapping search profiles. As shown in our evaluation on segmenting lung surfaces, this effectively solves the mesh folding problem and decreases the average absolute surface distance error from 0.82±0.29 mm (mean±standard deviation) to 0.79±0.24 mm.
KeywordsSegmentation Result Initial Vector Active Shape Model Gradient Vector Flow Surface Normal Direction
Unable to display preview. Download preview PDF.
- 5.Li, K., Jolly, M.: Simultaneous detection of multiple elastic surfaces with application to tumor segmentation in CT images. In: Proceedings of SPIE, San Diego, CA, USA, pp. 69143S–69143S–11 (2008)Google Scholar
- 6.Sun, S., McLennan, G., Hoffman, E.A., Beichel, R.: Model-based segmentation of pathological lungs in volumetric ct data. In: Proc. of Third International Workshop on Pulmonary Image Analysis, pp. 31–40 (2010)Google Scholar
- 8.Song, Q., Wu, X., Liu, Y., Smith, M., Buatti, J., Sonka, M.: Optimal graph search segmentation using Arc-Weighted graph for simultaneous surface detection of bladder and prostate. In: Yang, G.-Z., Hawkes, D., Rueckert, D., Noble, A., Taylor, C. (eds.) MICCAI 2009, Part II. LNCS, vol. 5762, pp. 827–835. Springer, Heidelberg (2009) ACM ID: 1691283CrossRefGoogle Scholar
- 9.Yin, Y., Zhang, X., Sonka, M.: Optimal multi-object multi-surface graph search segmentation: Full-joint cartilage delineation in 3d. In: Medical Image Understanding and Analysis, pp. 104–108 (2008)Google Scholar
- 11.Yin, Y.: Multi-surface, multi-object optimal image segmentation: application in 3D knee joint imaged by MR. PhD thesis, The University of Iowa (2010)Google Scholar