Paper Abstract and Keywords |
Presentation |
2016-01-20 09:56
Automated liver segmentation from 3D MRI without parameter tuning for imaging condition Yuto Masaki, Shunta Hirayama, Futoshi Yokota, Yoshito Otake (NAIST), Masatoshi Hori (Osaka Univ.), Toshiyuki Okada (Tsukuba Univ.), Noriyuki Tomiyama (Osaka Univ.), Yoshinobu Sato (NAIST) MI2015-114 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
The automated segmentation of liver from MRI is useful for computer-aided diagnosis system to liver fibrosis.
Several previous studies reported automated liver segmentation method from MRI, however their evaluation conducts to only a
specific imaging condition MRI. The automated liver segmentation method that does not depend on imaging modality and
condition is preferable. In our previous study, we reported automated liver segmentation method from CT using estimation of
target specific intensity model by shape-location priors. The purpose of this study is to evaluate liver segmentation accuracy
from MRI applied our method. We conducted experiments that our method was applied to contrast enhanced MRI and noncontrast
enhanced gradient echo MRI with parameters and training dataset obtained by CT. As a result, the dice coefficients were
0.955±0.020 and 0.934±0.025, respectively. Experimental results showed that our liver segmentation method in MRI was useful. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
subject-specific priors / bias field correction / statistical shape model / probabilistic atlas / joint histogram / computer-aided diagnosis / / |
Reference Info. |
IEICE Tech. Rep., vol. 115, no. 401, MI2015-114, pp. 199-204, Jan. 2016. |
Paper # |
MI2015-114 |
Date of Issue |
2016-01-12 (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 |
MI2015-114 |
|