Paper Abstract and Keywords |
Presentation |
2017-06-23 11:00
Similar CT image retrieval method based on lesion natures and their three-dimensional distribution Hiroaki Takebe, Yasutaka Moriwaki, Takayuki Baba (FUJITSU LAB.), Hiroaki Terada, Toru Higaki, Kazuo Awai (Hiroshima Univ.), Akio Ozawa, Kohei Murao, Yasuharu Ogino (FUJITSU), Mengjiao Wang, Rujie Liu (FRDC) PRMU2017-35 SP2017-11 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
When radiologists encounter a difficult case to diagnose, retrieval of morphologically similar cases in which final diagnosis is established may provide clinically useful information. Existing methods retrieve similar cases for each case in which lesions exist within a certain range of region by calculating similarity for the region specified on a slice image of the query case. However, radiologists diagnose cases of diffuse lung disease (DLD) where lesions appear throughout lungs by observing 3D distribution of lesion textures. We propose a novel method which can retrieve morphologically similar cases based on radiologist’s view for lesion textures and their 3-dimensional (3D) distribution. In our method, we configured a 3D model for central-peripheral regions of a lung and calculated similarity for 3D distribution of lesions using the 3D model. We evaluated the average precision of the proposed method for real CT images, and the experimental result showed that our method improved the accuracy compared with the comparative method. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Image diagnosis / CT image / Diffuse lung disease / Similar case retrieval / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 117, no. 105, PRMU2017-35, pp. 57-62, June 2017. |
Paper # |
PRMU2017-35 |
Date of Issue |
2017-06-15 (PRMU, SP) |
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 |
PRMU2017-35 SP2017-11 |
|