Paper Abstract and Keywords |
Presentation |
2018-09-21 10:00
[Short Paper]
Computer-Aided Diagnosis of Liver Cancers Using Deep Learning with Fine-tuning Weibin Wang (Ritsumeikan Univ.), Dong Liang, Lanfen Lin, Hongjie Hu, Qiaowei Zhang, Qingqing Chen (Zhejiang Univ.), Yutaro lwamoto, Xianhua Han, Yen-Wei Chen (Ritsumeikan Univ.) PRMU2018-57 IBISML2018-34 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Liver cancer is one of the leading causes of death world-wide. Computer-aided diagnosis plays an important role in liver lesion diagnosis (classification). Recently, several deep learning-based computer-aided diagnosis systems have been proposed for classification of liver lesions and their effectiveness have been demonstrated. The main challenge in deep learning-based medical image classification is the lack of annotated training samples. In this paper, we demonstrated that fine-tuning can significantly improve the liver lesion classification accuracy especially for the small training samples. We used the residual convolutional neural network (ResNet), which is the state-of-the-art network, as our baseline network for focal liver lesion classification on multi-phase CT images. The fine-tuning significantly improved the classification accuracy from 83.7% to 91.2%. The classification accuracy (91.2%) is higher than the accuracy of the state-of-the-art methods. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
ResNet / Liver cancer classification / Multi-phase CT / Fine-tuning / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 118, no. 219, PRMU2018-57, pp. 139-140, Sept. 2018. |
Paper # |
PRMU2018-57 |
Date of Issue |
2018-09-13 (PRMU, IBISML) |
ISSN |
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 |
PRMU2018-57 IBISML2018-34 |
|