Three Dimensional Concentration Index as a Feature of Line Pattern Distribution in Three Dimensional Space

Yasushi Hirano¹, Yoshito Mekada², Jun-ichi Hasegawa³ and Jun-ichiro Toriwaki⁴

¹Graduate School of Engineering, Nagoya University, Nagoya 464-8603, Japan
²Faculty of Engineering, Utsunomiya University, Utsunomiya 321-8585, Japan
³School of Computer and Cognitive Sciences, Chukyo University, Toyota 470-0393, Japan
⁴Faculty of Engineering, Nagoya University, Nagoya 464-8603, Japan

(Received July 21, 1998; Accepted February 3, 1999)

Keywords: 3D Concentration Index, Pattern Analysis, 3D Image Processing, Computer Aided Diagnosis, X-ray CT Image

Abstract. This paper proposes a new shape feature called three dimensional (3D) concentration index for analyzing local concentration of 3D line patterns. The 3D concentration index is conceptually an expansion of the 2D concentration index to 3D space. The 2D concentration index was used for detecting cancer lesions with local concentration of folds on the stomach wall in double contrast stomach X-ray images. In this paper, the basic properties of 3D concentration index are studied with simulations. Furthermore, we apply it to real chest X-ray CT images, and confirm its usefulness for the discrimination between benign and malignant lung tumor shadows.