Paper Abstract and Keywords |
Presentation |
2013-01-25 13:10
A pilot study of lung voxel classification for auto-detecting ground glass opacity nodules in chest CT images Mitsutaka Nemoto, Yoshitaka Masutani, Shouhei Hanaoka, Yukihiro Nomura, Soichiro Miki, Takeharu Yoshikawa, Naoto Hayashi, Kuni Ohtomo (Univ. Tokyo) MI2012-109 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In this study, we examine various voxel classification methods for auto-extracting ground glass opacity (GGO) nodule candidates. Classifiers evaluating experimentally are two types of one-class classifier (threshold functions of a Mahalanobis distance and a norm of residue by embedding in a manifold trained by Isomap), and two kinds of two-class classifier (a threshold function of Mahalanobis distance ratio and a classifier ensemble trained by AdaBoost). In addition to evaluate the classifiers individually, four cascade classifiers each of which are combination of an initial rejection by the one-class classifier and a following classification by the two-class classifier are also validated experimentally. The evaluation results show the cascade classifier has a potential for providing the voxel classification results to extract GGO nodule candidates with high performance. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Chest CT / ground glass opacity (GGO) nodule / computer aided detection (CAD / classifier / cascade / / / |
Reference Info. |
IEICE Tech. Rep., vol. 112, no. 411, MI2012-109, pp. 245-248, Jan. 2013. |
Paper # |
MI2012-109 |
Date of Issue |
2013-01-17 (MI) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
MI2012-109 |
|