Paper Abstract and Keywords |
Presentation |
2013-01-24 10:30
Construction of a sparse non-directional graphical model on anatomical landmark distances by the Graphical Lasso
-- Feasibility study for application to automatic landmark detection system -- Shouhei Hanaoka, Yoshitaka Masutani, Mitsutaka Nemoto, Yukihiro Nomura, Soichiro Miki, Takeharu Yoshikawa, Naoto Hayashi, Kuni Ohtomo (Univ. of Tokyo) MI2012-64 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We have been developed an automatic detection system for anatomical landmarks in CT images. The system utilizes a non-sparse multivariable Gaussian statistical model on inter-landmark distances. The aim of this study is to extend our previous method to sparse multivariable Gaussian model by GPGPU-implemented Graphical lasso method. The probabilistic distribution of inter-landmark distances was estimated as a multiple Gaussian distribution with a sparse precision matrix. Feasibility of improvement of the landmark detection accuracy with this novel inter-landmark distance model was evaluated. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Anatomical landmark / CT / statistical shape model / graphical lasso / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 112, no. 411, MI2012-64, pp. 13-18, Jan. 2013. |
Paper # |
MI2012-64 |
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 |
MI2012-64 |