Paper Abstract and Keywords |
Presentation |
2013-07-18 10:35
Comparison of sparse non-directional graphical models on anatomical landmark distances by the Smoothly Clipped Absolute Deviation (SCAD) and the Graphical Lasso Shouhei Hanaoka (Univ. of Tokyo Hospital), Yoshitaka Masutani (Univ. of Tokyo Hospital/Univ. of Tokyo), Mitsutaka Nemoto, Yukihiro Nomura, Soichiro Miki, Takeharu Yoshikawa, Naoto Hayashi (Univ. of Tokyo Hospital), Kuni Ohtomo (Univ. of Tokyo Hospital/Univ. of Tokyo) MI2013-20 |
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 multivariable Gaussian statistical model on inter-landmark distances which were estimated with Tikhonov or L1-norm regularization terms. The aim of this study is to introduce a novel SCAD regularization term to our landmark detection system. In the experimental result, the SCAD-based method showed almost the same result as L1 norm-based method. However, the former showed more robustness against parameter change than the latter method |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
anatomical landmark / CT / statistical shape odel / graphical lasso / SCAD / / / |
Reference Info. |
IEICE Tech. Rep., vol. 113, no. 146, MI2013-20, pp. 7-12, July 2013. |
Paper # |
MI2013-20 |
Date of Issue |
2013-07-11 (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 |
MI2013-20 |
|