Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology

Hiroki Tokunaga, Brian Kenji Iwana, Yuki Teramoto, Akihiko Yoshizawa, Ryoma Bise

研究成果: Chapter in Book/Report/Conference proceedingConference contribution

抄録

In pathological diagnosis, since the proportion of the adenocarcinoma subtypes is related to the recurrence rate and the survival time after surgery, the proportion of cancer subtypes for pathological images has been recorded as diagnostic information in some hospitals. In this paper, we propose a subtype segmentation method that uses such proportional labels as weakly supervised labels. If the estimated class rate is higher than that of the annotated class rate, we generate negative pseudo labels, which indicate, “input image does not belong to this negative label,” in addition to standard pseudo labels. It can force out the low confidence samples and mitigate the problem of positive pseudo label learning which cannot label low confident unlabeled samples. Our method outperformed the state-of-the-art semi-supervised learning (SSL) methods.

本文言語英語
ホスト出版物のタイトルComputer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings
編集者Andrea Vedaldi, Horst Bischof, Thomas Brox, Jan-Michael Frahm
出版社Springer Science and Business Media Deutschland GmbH
ページ430-446
ページ数17
ISBN(印刷版)9783030585549
DOI
出版ステータス出版済み - 2020
イベント16th European Conference on Computer Vision, ECCV 2020 - Glasgow, 英国
継続期間: 8 23 20208 28 2020

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
12360 LNCS
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

会議

会議16th European Conference on Computer Vision, ECCV 2020
国/地域英国
CityGlasgow
Period8/23/208/28/20

All Science Journal Classification (ASJC) codes

  • 理論的コンピュータサイエンス
  • コンピュータ サイエンス(全般)

フィンガープリント

「Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル