TY - GEN
T1 - Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology
AU - Tokunaga, Hiroki
AU - Iwana, Brian Kenji
AU - Teramoto, Yuki
AU - Yoshizawa, Akihiko
AU - Bise, Ryoma
N1 - Funding Information:
This work was supported by JSPS KAKENHI Grant Number 20H04211.
Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - 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.
AB - 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.
UR - http://www.scopus.com/inward/record.url?scp=85097372683&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85097372683&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-58555-6_26
DO - 10.1007/978-3-030-58555-6_26
M3 - Conference contribution
AN - SCOPUS:85097372683
SN - 9783030585549
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 430
EP - 446
BT - Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings
A2 - Vedaldi, Andrea
A2 - Bischof, Horst
A2 - Brox, Thomas
A2 - Frahm, Jan-Michael
PB - Springer Science and Business Media Deutschland GmbH
T2 - 16th European Conference on Computer Vision, ECCV 2020
Y2 - 23 August 2020 through 28 August 2020
ER -