Translator Disclaimer
9 May 2002 CAD system for lung cancer based on low-dose single-slice CT image
Author Affiliations +
We have been developed a computer-aided diagnosis (CAD) system for the lung cancer detection of early stage from low dose single-slice computed tomography (CT) with 10 mm beam width on chest screening. The objective of this study is to solve three problems of the conventional CAD system; (1) lesion which overlaps blood vessel, (2) lesion in contact with blood vessel and (3) lesion near upper mediastinum. This paper presents a new method to solve problem-1 and problem-2. The blood vessels, which overlap lesions and others in contact with lesion, are eliminated by detecting region of interest (ROI) with accuracy. Detection method of ROIs consists of 3 processes; firstly, streak shadows elimination using linear feature detector filter, secondly, estimation of pulmonary background bias using the intensity histogram and the opening method, and finally, ROI's border detection using laplacian filter. We evaluated the new system by apply it to 155 shadows which need confirmation diagnosis. These cases were selected from clinical test from July 1997 to December 2000 in retrospective study. True positive cases of this algorithm achieved sensitivity 91.0 %. The average of false positive cases was 0.53 per slice.
© (2002) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mitsuru Kubo, Kazuhori Kubota, Nobuhiro Yamada, Yoshiki Kawata, Noboru Niki, Kenji Eguchi, Hironobu Ohmatsu, Ryutaro Kakinuma, Masahiro Kaneko, Masahiko Kusumoto, Kiyoshi Mori, Hiroyuki Nishiyama, and Noriyuki Moriyama "CAD system for lung cancer based on low-dose single-slice CT image", Proc. SPIE 4684, Medical Imaging 2002: Image Processing, (9 May 2002);

Back to Top