Paper Abstract and Keywords |
Presentation |
2020-01-29 13:20
[Short Paper]
Automatic segmentation of malignant tumors using PET/CT images and statistical images Manami Haga, Takeshi Hara (Gifu Univ.), Satoshi Ito, Masaya Kato (Daiyukai Hospital), Masaki Matsusako (St.Luke's Hospital), Zhou Xiangrong (Gifu Univ.), Tetsuro Katafuchi (Gifu Univ. of Medical Science), Hiroshi Fujita (Gifu Univ.) MI2019-83 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
The purpose of this study was to develop an automated detection system of tumors in FDG-PET/CT images. In this work, an anatomical standardization process was performed based on landmarks on organs’ surfaces determined by semi-automated GraphCut method on CT images. After the anatomical standardization of FDG-PET images, the Z-score images in each patient were obtained from the mean and the standard deviation images. Suspicious regions on FDG-PET images were determined using a dynamic thresholding approach and labeling methods. An convolutional neural network was used to discriminate the remaining regions as false-positive areas. Detection performance was evaluated 65 PET/CT cases. The performance was 79.8% sensitivity with 22.7 marks per case. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
PET / SUV / Z-score / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 119, no. 399, MI2019-83, pp. 77-81, Jan. 2020. |
Paper # |
MI2019-83 |
Date of Issue |
2020-01-22 (MI) |
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 |
MI2019-83 |
|