半監督語義分割模型的有效性在很大程度上取決於作為可信偽標簽的高置信度像素的生成和利用。然而，當用於生成這些偽標簽的推理模型訓練不足時，就會出現挑戰，導致偽標簽盡管可信度很高，但卻不可靠。為解決這一問題，本研究引入了可靠性評估模塊，通過使用一組精心挑選的模型來評估候選偽標簽的可靠性，從而改進可靠偽標簽的選擇。此外，還提出了一種整合了各種增強技術的新型自我訓練方法，稱作“可靠偽標簽的一致性訓練和選擇（CTSR）”的框架。在三個開放的醫學數據集上進行的綜合實驗表明，在分割準確性方面明顯優於現有方法。

The effectiveness of semi-supervised semantic segmentation models depends heavily on the generation and utilization of high-confidence pixels as trusted pseudo-labels. However, challenges arise when the inference models used to generate these pseudo labels are under-trained, resulting in labels that are unreliable despite their high confidence scores. To address this problem, this study introduces a reliability assessment module, which improves the selection of reliable pseudo-labels by evaluating the reliability of the candidate pseudo-labels using a set of carefully selected models. In addition, a novel self-training method is proposed that integrates various augmentation techniques. Comprehensive experiments on three open medical datasets demonstrate that the proposed framework, designated as Consistent Training and Selection of Reliable Pseudo Labels (CTSR), significantly outperforms existing methods in terms of the segmentation accuracy.

[1] Lihe Yang, Lei Qi, Litong Feng, Wayne Zhang, and Yinghuan Shi. Revisiting weak-to-strong consistency in semi-supervised semantic segmentation. In CVPR, 2023.
[2] Yuchao Wang, Haochen Wang, Yujun Shen, Jingjing Fei, Wei Li, Guoqiang Jin, Liwei Wu, Rui Zhao, and Xinyi Le. Semi-supervised semantic segmentation using unreliable pseudo-labels, 2022.
[3] Yude Wang, Jie Zhang, Meina Kan, and Shiguang Shan. Learning pseudo labels for semi-and-weakly supervised semantic segmentation. Pattern Recognition, 132:108925, 2022.
[4] Zhen Zhao, Lihe Yang, Sifan Long, Jimin Pi, Luping Zhou, and Jingdong Wang. Aug-mentation matters: A simple-yet-effective approach to semi-supervised semantic seg-mentation, 2022.
[5] Xiaokang Chen, Yuhui Yuan, Gang Zeng, and Jingdong Wang. Semi-supervised se-mantic segmentation with cross pseudo supervision, 2021.
[6] Eric Arazo, Diego Ortego, Paul Albert, Noel E O’Connor, and Kevin McGuinness. Pseudo-labeling and confirmation bias in deep semi-supervised learning. In 2020 In-ternational joint conference on neural networks (IJCNN), pages 1–8. IEEE, 2020.
[7] Sotiris B Kotsiantis, Ioannis Zaharakis, P Pintelas, et al. Supervised machine learning: A review of classification techniques. Emerging artificial intelligence applications in computer engineering, 160(1):3–24, 2007.
[8] Yves Grandvalet and Yoshua Bengio. Semi-supervised learning by entropy minimiza-tion. Advances in neural information processing systems, 17, 2004.
[9] Xiaojin Zhu and Andrew B Goldberg. Introduction to semi-supervised learning. Springer Nature, 2022.
[10] Zongwei Zhou, Md Mahfuzur Rahman Siddiquee, Nima Tajbakhsh, and Jianming Liang. Unet++: A nested u-net architecture for medical image segmentation. In Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support: 4th International Workshop, DLMIA 2018, and 8th International Workshop, ML-CDS 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings 4, pages 3–11. Springer, 2018.
[11] Yu Weng, Tianbao Zhou, Yujie Li, and Xiaoyu Qiu. Nas-unet: Neural architecture search for medical image segmentation. IEEE access, 7:44247–44257, 2019.
[12] Fausto Milletari, Nassir Navab, and Seyed-Ahmad Ahmadi. V-net: Fully convolutional neural networks for volumetric medical image segmentation. In 2016 fourth interna-tional conference on 3D vision (3DV), pages 565–571. Ieee, 2016.
[13] Hu Cao, Yueyue Wang, Joy Chen, Dongsheng Jiang, Xiaopeng Zhang, Qi Tian, and Manning Wang. Swin-unet: Unet-like pure transformer for medical image segmenta-tion. In European conference on computer vision, pages 205–218. Springer, 2022.
[14] Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted win-dows. In Proceedings of the IEEE/CVF international conference on computer vision, pages 10012–10022, 2021.
[15] Jieneng Chen, Yongyi Lu, Qihang Yu, Xiangde Luo, Ehsan Adeli, Yan Wang, Le Lu, Alan L Yuille, and Yuyin Zhou. Transunet: Transformers make strong encoders for medical image segmentation. arXiv preprint arXiv:2102.04306, 2021.
[16] Debesh Jha, Pia H Smedsrud, Michael A Riegler, Dag Johansen, Thomas De Lange, P˚al Halvorsen, and H˚avard D Johansen. Resunet++: An advanced architecture for medical image segmentation. In 2019 IEEE international symposium on multimedia (ISM), pages 225–2255. IEEE, 2019.
[17] Juntang Zhuang. Laddernet: Multi-path networks based on u-net for medical image segmentation. arXiv preprint arXiv:1810.07810, 2018.
[18] Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and efficient design for semantic segmentation with trans-formers. Advances in neural information processing systems, 34:12077–12090, 2021.
[19] Kaiping Wang, Bo Zhan, Chen Zu, Xi Wu, Jiliu Zhou, Luping Zhou, and Yan Wang. Semi-supervised medical image segmentation via a tripled-uncertainty guided mean teacher model with contrastive learning. Medical Image Analysis, 79:102447, 2022.
[20] Kuo-Sheng Cheng, Qiong-Wen Zhang, Hung-Wen Tsai, Nien-tsu Li, and Pau-Choo Chung. Domain-centroid-guided progressive teacher-based knowledge distillation for source-free domain adaptation of histopathological images. IEEE Transactions on Ar-tificial Intelligence, pages 1–14, 2023.
[21] Seng Hansun. A new approach of moving average method in time series analysis. In 2013 conference on new media studies (CoNMedia), pages 1–4. IEEE, 2013.
[22] Rahaf Aljundi, Francesca Babiloni, Mohamed Elhoseiny, Marcus Rohrbach, and Tinne Tuytelaars. Memory aware synapses: Learning what (not) to forget. In Proceedings of the European conference on computer vision (ECCV), pages 139–154, 2018.
[23] Yves Grandvalet and Yoshua Bengio. Semi-supervised learning by entropy minimiza-tion. Advances in neural information processing systems, 17, 2004.
[24] Kihyuk Sohn, David Berthelot, Nicholas Carlini, Zizhao Zhang, Han Zhang, Colin A Raffel, Ekin Dogus Cubuk, Alexey Kurakin, and Chun-Liang Li. Fixmatch: Simpli-fying semi-supervised learning with consistency and confidence. Advances in neural information processing systems, 33:596–608, 2020.
[25] Sudhanshu Mittal, Maxim Tatarchenko, and Thomas Brox. Semi-supervised seman-tic segmentation with high- and low-level consistency. IEEE Transactions on Pattern Analysis and Machine Intelligence, 43(4):1369–1379, 2021.
[26] Yang Chen, Weiran Wang, and Chao Wang. Semi-supervised asr by end-to-end self-training. arXiv preprint arXiv:2001.09128, 2020.
[27] Massih-Reza Amini, Vasilii Feofanov, Loic Pauletto, Emilie Devijver, and Yury Maxi-mov. Self-training: A survey. arXiv preprint arXiv:2202.12040, 2022.
[28] Fabio Roli and Gian Luca Marcialis. Semi-supervised pca-based face recognition using self-training. In Structural, Syntactic, and Statistical Pattern Recognition: Joint IAPR International Workshops, SSPR 2006 and SPR 2006, Hong Kong, China, August 17-19, 2006. Proceedings, pages 560–568. Springer, 2006.
[29] Chen Wei, Kihyuk Sohn, Clayton Mellina, Alan Yuille, and Fan Yang. Crest: A class-rebalancing self-training framework for imbalanced semi-supervised learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10857–10866, 2021.
[30] Qizhe Xie, Minh-Thang Luong, Eduard Hovy, and Quoc V Le. Self-training with noisy student improves imagenet classification. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10687–10698, 2020.
[31] Lihe Yang, Wei Zhuo, Lei Qi, Yinghuan Shi, and Yang Gao. St++: Make self-training work better for semi-supervised semantic segmentation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4268–4277, 2022.
[32] Carole H Sudre, Wenqi Li, Tom Vercauteren, Sebastien Ourselin, and M Jorge Car-doso. Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations. In Deep Learning in Medical Image Analysis and Multimodal Learn-ing for Clinical Decision Support: Third International Workshop, DLMIA 2017, and 7th International Workshop, ML-CDS 2017, Held in Conjunction with MICCAI 2017, Qu´ebec City, QC, Canada, September 14, Proceedings 3, pages 240–248. Springer, 2017.
[33] Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Doll´ar. Focal loss for dense object detection. In Proceedings of the IEEE international conference on computer vision, pages 2980–2988, 2017.
[34] Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016.
[35] Wu et al. Quantification of oil droplets for the diagnosis of nonalcoholic fatty liver disease based on regional and borderline probability of case segmentation. PhD thesis, 2021.
[36] Caglar Senaras, M Khalid Khan Niazi, Gerard Lozanski, and Metin N Gurcan. Deepfo-cus: detection of out-of-focus regions in whole slide digital images using deep learning. PloS one, 13(10):e0205387, 2018.
[37] Debesh Jha, Pia H. Smedsrud, Michael A. Riegler, P˚al Halvorsen, Thomas de Lange, Dag Johansen, and H˚avard D. Johansen. Kvasir-seg: A segmented polyp dataset, 2019.
[38] Xinkai Zhao, Chaowei Fang, De-Jun Fan, Xutao Lin, Feng Gao, and Guanbin Li. Cross-level contrastive learning and consistency constraint for semi-supervised med-ical image segmentation, 2022.
[39] Noel Codella, Veronica Rotemberg, Philipp Tschandl, M Emre Celebi, Stephen Dusza, David Gutman, Brian Helba, Aadi Kalloo, Konstantinos Liopyris, Michael Marchetti, et al. Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (isic). arXiv preprint arXiv:1902.03368, 2019.
[40] Zhiqiang Shen, Peng Cao, Hua Yang, Xiaoli Liu, Jinzhu Yang, and Osmar R Zaiane. Co-training with high-confidence pseudo labels for semi-supervised medical image segmentation. arXiv preprint arXiv:2301.04465, 2023.
[41] Antti Tarvainen and Harri Valpola. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. Advances in neural information processing systems, 30, 2017.
[42] Yassine Ouali, C´eline Hudelot, and Myriam Tami. Semi-supervised semantic segmen-tation with cross-consistency training. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 12674–12684, 2020.
[43] Tao Wang, Jianglin Lu, Zhihui Lai, Jiajun Wen, and Heng Kong. Uncertainty-guided pixel contrastive learning for semi-supervised medical image segmentation. In Inter-national Joint Conference on Artificial Intelligence, 2022.