Automatic cell analysis for P53 immunohistochemistry in bladder inverted papilloma

Author

Zhao, P. ; Mao, K.Z. ; Koh, T.S. ; Tan, P.H.

Author_Institution

Sch. of Electr. & Electron. Eng., Nanyang Technol. Univ., Singapore

fYear

2003

fDate

20-22 Oct. 2003

Firstpage

168

Lastpage

169

Abstract

The automatic cell analysis system developed in the present study has two merits. Firstly, a nonlinear hybrid color space is built for pixel classification. The components of the nonlinear hybrid color space are selected from RGB, HSI, XYZ , L´ u´ v´, and nonlinear polynomial combinations of these components, with the objective of achieving large class separability between nuclei and non-nuclei pixels. Secondly, a novel segmentation algorithm for overlapping or touching nuclei is developed by incorporating hierarchical clustering into distance transform to locate nuclei centers. The effectiveness of the proposed methods is tested using a few images.

Keywords

biochemistry; biological organs; biological tissues; cellular biophysics; image classification; image segmentation; medical image processing; molecular biophysics; pattern clustering; proteins; HSI color space; L´ u´ v´ color space; P53 immunohistochemistry; RGB color space; XYZ color space; automatic cell analysis; bladder inverted papilloma; class separability; distance transform; hierarchical clustering; nonlinear hybrid color space; nonlinear polynomial combinations; nonnuclei pixels; nuclei pixels; pixel classification; protein; segmentation algorithm; touching nuclei; Bladder; Hospitals; Image analysis; Image segmentation; Immune system; Pathology; Performance evaluation; Pixel; Polynomials; Space technology;

fLanguage

English

Publisher

ieee

Conference_Titel

Biomedical Engineering, 2003. IEEE EMBS Asian-Pacific Conference on

Print_ISBN

0-7803-7943-8

Type

conf

DOI

10.1109/APBME.2003.1302637

Filename

1302637