Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology
Abstract
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.
Cite
Text
Tokunaga et al. "Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology." Proceedings of the European Conference on Computer Vision (ECCV), 2020. doi:10.1007/978-3-030-58555-6_26Markdown
[Tokunaga et al. "Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology." Proceedings of the European Conference on Computer Vision (ECCV), 2020.](https://mlanthology.org/eccv/2020/tokunaga2020eccv-negative/) doi:10.1007/978-3-030-58555-6_26BibTeX
@inproceedings{tokunaga2020eccv-negative,
title = {{Negative Pseudo Labeling Using Class Proportion for Semantic Segmentation in Pathology}},
author = {Tokunaga, Hiroki and Iwana, Brian Kenji and Teramoto, Yuki and Yoshizawa, Akihiko and Bise, Ryoma},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},
year = {2020},
doi = {10.1007/978-3-030-58555-6_26},
url = {https://mlanthology.org/eccv/2020/tokunaga2020eccv-negative/}
}