Paper Abstract and Keywords |
Presentation |
2017-11-07 16:00
[Invited Talk]
Application of Real-time Image Recognition System with Machine and Transfer Learnings to Computer-Aided Diagnosis for Endoscopic Images of Colorectal Cancer Tetsushi Koide, Toru Tamaki (Hiroshima Univ.), Shigeto Yoshida, Hiroshi Mieno (Medical Corp. JR Hiroshima Hospital), Shinji Tanaka (Hiroshima Univ. Hospital) CPSY2017-42 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We have developed a real-time computer-aided diagnosis systems for colorectal endoscopic images using machine, transfer, and deep learnings, and it learns clinical doctor’s the NBI magnification findings of colorectal endoscopic images. The developed system can be identified the colorectal endoscopic images either non-neoplastic or neoplastic lesions in real time. We can provide the several software / hardware solutions according to the clinical demands.
The accuracy calculated for evaluating concordance between diagnosis by the system and the histological findings was 94.9% (112/118). Further, the concordance between the endoscopic diagnosis and diagnosis by the system was 97.5% (115/118) (κ=0.94).For the adenomatous histology of lesions (polyps) < 5 mm, the accuracy between the histological findings of diminutive colorectal lesions and the diagnosis by the system was 93.2% (82/88).The concordance between the endoscopic diagnosis and diagnosis by the system was 96,6% (85/88) (κ=0.93), and the agreement was excellent. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Diagnosis for Colorectal and Gastrointestinal Endoscopic Images / Computer-Aided Diagnosis (CAD) System / Real-Time Image Recognition / Machine Learning / Transfer Learning / Deep Learning / Navigation / Dedicated Hardware |
Reference Info. |
IEICE Tech. Rep., vol. 117, no. 278, CPSY2017-42, pp. 19-19, Nov. 2017. |
Paper # |
CPSY2017-42 |
Date of Issue |
2017-10-31 (CPSY) |
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 |
CPSY2017-42 |