MT2: Multi-task Mean Teacher for Semi-Supervised Cell Segmentation

Binyu Zhang, Junhao Dong, Zhicheng Zhao, Zhu Meng, Fei Su
Proceedings of The Cell Segmentation Challenge in Multi-modality High-Resolution Microscopy Images, PMLR 212:1-13, 2023.

Abstract

Cell segmentation is significant for downstream single-cell analysis in biological and biomedical research. Recently, image segmentation methods based on supervised learning have achieved promising results. However, most of them rely on intensive manual annotations, which are extremely time-consuming and expensive for cell segmentation. In addition, existing methods are often trained for a specific modality with poor generalization ability. In this paper, a novel semi-supervised cell segmentation method is proposed to segment microscopy images from multiple modalities. Specifically, Mean Teacher model is introduced to a multi-task learning framework, named Multi-task Mean Teacher (MT$^{2}$), in which both the classification and the regression heads are utilized to improve the prediction performance. Moreover, new data augmentation and multi-scale inference strategies are presented to enhance the robustness and generalization ability. For the quantitative evaluation on the Tuning Set of NeurIPS 2022 Cell Segmentation Competition, our method achieves the F1 Score of 0.8690, which demonstrates the effectiveness of the proposed semi-supervised learning method. Code is available at \url{https://github.com/djh-dzxw/MT2}

Cite this Paper

BibTeX
@InProceedings{pmlr-v212-zhang23a, title = {MT2: Multi-task Mean Teacher for Semi-Supervised Cell Segmentation}, author = {Zhang, Binyu and Dong, Junhao and Zhao, Zhicheng and Meng, Zhu and Su, Fei}, booktitle = {Proceedings of The Cell Segmentation Challenge in Multi-modality High-Resolution Microscopy Images}, pages = {1--13}, year = {2023}, editor = {Ma, Jun and Xie, Ronald and Gupta, Anubha and Guilherme de Almeida, José and Bader, Gary D. and Wang, Bo}, volume = {212}, series = {Proceedings of Machine Learning Research}, month = {28 Nov--09 Dec}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v212/zhang23a/zhang23a.pdf}, url = {https://proceedings.mlr.press/v212/zhang23a.html}, abstract = {Cell segmentation is significant for downstream single-cell analysis in biological and biomedical research. Recently, image segmentation methods based on supervised learning have achieved promising results. However, most of them rely on intensive manual annotations, which are extremely time-consuming and expensive for cell segmentation. In addition, existing methods are often trained for a specific modality with poor generalization ability. In this paper, a novel semi-supervised cell segmentation method is proposed to segment microscopy images from multiple modalities. Specifically, Mean Teacher model is introduced to a multi-task learning framework, named Multi-task Mean Teacher (MT$^{2}$), in which both the classification and the regression heads are utilized to improve the prediction performance. Moreover, new data augmentation and multi-scale inference strategies are presented to enhance the robustness and generalization ability. For the quantitative evaluation on the Tuning Set of NeurIPS 2022 Cell Segmentation Competition, our method achieves the F1 Score of 0.8690, which demonstrates the effectiveness of the proposed semi-supervised learning method. Code is available at \url{https://github.com/djh-dzxw/MT2}} }
Endnote
%0 Conference Paper %T MT2: Multi-task Mean Teacher for Semi-Supervised Cell Segmentation %A Binyu Zhang %A Junhao Dong %A Zhicheng Zhao %A Zhu Meng %A Fei Su %B Proceedings of The Cell Segmentation Challenge in Multi-modality High-Resolution Microscopy Images %C Proceedings of Machine Learning Research %D 2023 %E Jun Ma %E Ronald Xie %E Anubha Gupta %E José Guilherme de Almeida %E Gary D. Bader %E Bo Wang %F pmlr-v212-zhang23a %I PMLR %P 1--13 %U https://proceedings.mlr.press/v212/zhang23a.html %V 212 %X Cell segmentation is significant for downstream single-cell analysis in biological and biomedical research. Recently, image segmentation methods based on supervised learning have achieved promising results. However, most of them rely on intensive manual annotations, which are extremely time-consuming and expensive for cell segmentation. In addition, existing methods are often trained for a specific modality with poor generalization ability. In this paper, a novel semi-supervised cell segmentation method is proposed to segment microscopy images from multiple modalities. Specifically, Mean Teacher model is introduced to a multi-task learning framework, named Multi-task Mean Teacher (MT$^{2}$), in which both the classification and the regression heads are utilized to improve the prediction performance. Moreover, new data augmentation and multi-scale inference strategies are presented to enhance the robustness and generalization ability. For the quantitative evaluation on the Tuning Set of NeurIPS 2022 Cell Segmentation Competition, our method achieves the F1 Score of 0.8690, which demonstrates the effectiveness of the proposed semi-supervised learning method. Code is available at \url{https://github.com/djh-dzxw/MT2}
APA
Zhang, B., Dong, J., Zhao, Z., Meng, Z. & Su, F.. (2023). MT2: Multi-task Mean Teacher for Semi-Supervised Cell Segmentation. Proceedings of The Cell Segmentation Challenge in Multi-modality High-Resolution Microscopy Images, in Proceedings of Machine Learning Research 212:1-13 Available from https://proceedings.mlr.press/v212/zhang23a.html.

Related Material