Paper Abstract and Keywords |
Presentation |
2011-05-19 11:00
A System for Colorectal Endoscopic Images based on NBI Magnification Findings Junki Yoshimuta, Toru Tamaki, Bisser Raytchev, Kazufumi Kaneda, Yoshito Takemura, Shigeto Yoshida, Shinji Tanaka (Hiroshima Univ.) IE2011-11 PRMU2011-3 MI2011-3 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Our research objective is to classify magnified endoscopic images of colorectal tumours into 3 types (Type A, B, and C3) based on NBI magnification findings.
In this technical report, we propose a system for classifying colorectal endoscopic images.
These types are classified by pit and microvessel structure of colorectal mucosa, and have a strong correlation with histologic diagnosis: hyperplasias (HP), tubular adenomas (TA), and carcinomas with massive submucosal invasion (sm-m).
A dataset of 908 NBI images are classified by using a SVM classifier with densely sampled SIFT descriptors, and a rate of 96.00% with 10-fold cross validation was achieved. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Medical Imaging / Pattern Recognition / Dense SIFT / Bag-of-features / SVM / NBI / Colorectal Cancer / |
Reference Info. |
IEICE Tech. Rep., vol. 111, no. 48, PRMU2011-3, pp. 13-18, May 2011. |
Paper # |
PRMU2011-3 |
Date of Issue |
2011-05-12 (IE, PRMU, 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 |
IE2011-11 PRMU2011-3 MI2011-3 |
|