Image Segmentation and Analysis Method for 3- or Higher-dimenstional Images
This system and method identifies optimal surface and boundary regions in 3- or higher-dimentional images containg mutiple intersecting objects.
UIRF Case #:04064
|Relevant Publications||Level of Development|
|Technology Description||Inventor Web Site Link|
|Patent Links||Contact Information|
Li K, Wu X, Chen DZ, Sonka M. Optimal Surface Segmentation in Volumetric Images�A Graph-Theoretic Approach. IEEE Trans. Pattern Analysis Machine Intell. 2006; 28(1): 119 � 134.
Level of Development
General: Advanced Prototype
Researchers at University of Iowa have developed a graph-based n-dimensional image segmentation method that can be used directly on three or higher dimensional images. The same principle can be applied to any n-dimensional data sets containing non-image information. An image as well as any scalar or vector function in n-dimensional space can be represented by series of individual coordinates having measureable properties that can be analyzed by the series of computational algorithms to determine whether a set of coordinates are on a continuous line and/or surface. The method validation has been completed by using volumetric medical images from CT, MR, and ultrasound, and the results demonstrated accurate and simultaneous detection of multiple intersecting edges and surfaces. This method also provides significant improvements to the previous attempts on expanding the graph-searching segmentation techniques to analysis of data sets with higher dimensions.
Inventor Web Site Link(s)