DocumentCode
461946
Title
Hemispherical Harmonic Surface Description and Applications to Medical Image Analysis
Author
Huang, Heng ; Zhang, Lei ; Samaras, Dimitris ; Shen, Li ; Zhang, Rong ; Makedon, Fillia ; Pearlman, Justin
Author_Institution
Dept. of Comput. Sci., Dartmouth Coll., Hanover, NH
fYear
2006
fDate
14-16 June 2006
Firstpage
381
Lastpage
388
Abstract
The use of surface harmonics for rigid and nonrigid shape description is well known. In this paper we define a set of complete hemispherical harmonic basis functions on a hemisphere domain and propose a novel parametric shape description method to efficiently and flexibly represent the surfaces of anatomical structures in medical images. As the first application of hemispherical harmonic theory in shape description, our technique differs from the previous surface harmonics shape descriptors, all of which don´t work efficiently on the hemisphere-like objects that often exist in medical anatomical structures (e.g., ventricles, atriums, etc.). We demonstrate the effectiveness of our approach through theoretic and experimental exploration of a set of medical image applications. Furthermore, an evaluation criterion for surface modeling efficiency is described and the comparison results demonstrated that our method outperformed the previous approaches using spherical harmonic models.
Keywords
harmonic analysis; medical image processing; surface fitting; hemispherical harmonic basis function; hemispherical harmonic surface description; medical image analysis; nonrigid shape description; parametric shape description method; Anatomical structure; Application software; Biomedical imaging; Computer science; Computer vision; Image analysis; Medical diagnostic imaging; Shape; Surface morphology; Surface reconstruction;
fLanguage
English
Publisher
ieee
Conference_Titel
3D Data Processing, Visualization, and Transmission, Third International Symposium on
Conference_Location
Chapel Hill, NC
Print_ISBN
0-7695-2825-2
Type
conf
DOI
10.1109/3DPVT.2006.71
Filename
4155751
Link To Document