CASC-AI: Consensus-aware Self-corrective Learning for Cell Segmentation with Noisy Labels

Ruining Deng, Yihe Yang, David J Pisapia, Benjamin L Liechty, Junchao Zhu, Juming Xiong, Junlin Guo, Zhengyi Lu, Jiacheng Wang, Xing Yao, Runxuan Yu, Rendong Zhang, Gaurav Rudravaram, Mengmeng Yin, Pinaki Sarder, Haichun Yang, Yuankai Huo, Mert R. Sabuncu
Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, PMLR 301:293-309, 2026.

Abstract

Multi-class cell segmentation in high-resolution gigapixel whole slide images (WSIs) is crucial for various clinical applications. However, training such models typically requires labor-intensive, pixel-wise annotations by domain experts. Recent efforts have democratized this process by involving lay annotators without medical expertise. However, conventional non-corrective approaches struggle to handle annotation noise adaptively because they lack mechanisms to mitigate false positives (FP) and false negatives (FN) at both the image-feature and pixel levels. In this paper, we propose a consensus-aware self-corrective learning that leverages the Consensus Matrix to guide its learning process. The Consensus Matrix defines regions where both the AI and annotators agree on cell and non-cell annotations, which are prioritized with stronger supervision. Conversely, areas of disagreement are adaptively weighted based on their feature similarity to high-confidence consensus regions, with more similar regions receiving greater attention. Additionally, contrastive learning is employed to separate features of noisy regions from those of reliable consensus regions by maximizing their dissimilarity. This paradigm enables the model to iteratively refine noisy labels, enhancing its robustness. Validated on one real-world lay-annotated cell dataset and two reasoning-guided simulated noisy datasets, our method demonstrates improved segmentation performance, effectively correcting FP and FN errors and showcasing its potential for training robust models on noisy datasets. The official implementation and cell annotations are publicly available at https://github.com/ddrrnn123/CASC-AI.

Cite this Paper

BibTeX
@InProceedings{pmlr-v301-deng26a, title = {CASC-AI: Consensus-aware Self-corrective Learning for Cell Segmentation with Noisy Labels}, author = {Deng, Ruining and Yang, Yihe and Pisapia, David J and Liechty, Benjamin L and Zhu, Junchao and Xiong, Juming and Guo, Junlin and Lu, Zhengyi and Wang, Jiacheng and Yao, Xing and Yu, Runxuan and Zhang, Rendong and Rudravaram, Gaurav and Yin, Mengmeng and Sarder, Pinaki and Yang, Haichun and Huo, Yuankai and Sabuncu, Mert R.}, booktitle = {Proceedings of The 8th International Conference on Medical Imaging with Deep Learning}, pages = {293--309}, year = {2026}, editor = {Tasdizen, Tolga and Elhabian, Shireen and Summers, Ronald and Chen, Chen and Koch, Lisa and Zhuang, Yan}, volume = {301}, series = {Proceedings of Machine Learning Research}, month = {09--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v301/main/assets/deng26a/deng26a.pdf}, url = {https://proceedings.mlr.press/v301/deng26a.html}, abstract = {Multi-class cell segmentation in high-resolution gigapixel whole slide images (WSIs) is crucial for various clinical applications. However, training such models typically requires labor-intensive, pixel-wise annotations by domain experts. Recent efforts have democratized this process by involving lay annotators without medical expertise. However, conventional non-corrective approaches struggle to handle annotation noise adaptively because they lack mechanisms to mitigate false positives (FP) and false negatives (FN) at both the image-feature and pixel levels. In this paper, we propose a consensus-aware self-corrective learning that leverages the Consensus Matrix to guide its learning process. The Consensus Matrix defines regions where both the AI and annotators agree on cell and non-cell annotations, which are prioritized with stronger supervision. Conversely, areas of disagreement are adaptively weighted based on their feature similarity to high-confidence consensus regions, with more similar regions receiving greater attention. Additionally, contrastive learning is employed to separate features of noisy regions from those of reliable consensus regions by maximizing their dissimilarity. This paradigm enables the model to iteratively refine noisy labels, enhancing its robustness. Validated on one real-world lay-annotated cell dataset and two reasoning-guided simulated noisy datasets, our method demonstrates improved segmentation performance, effectively correcting FP and FN errors and showcasing its potential for training robust models on noisy datasets. The official implementation and cell annotations are publicly available at https://github.com/ddrrnn123/CASC-AI.} }
Endnote
%0 Conference Paper %T CASC-AI: Consensus-aware Self-corrective Learning for Cell Segmentation with Noisy Labels %A Ruining Deng %A Yihe Yang %A David J Pisapia %A Benjamin L Liechty %A Junchao Zhu %A Juming Xiong %A Junlin Guo %A Zhengyi Lu %A Jiacheng Wang %A Xing Yao %A Runxuan Yu %A Rendong Zhang %A Gaurav Rudravaram %A Mengmeng Yin %A Pinaki Sarder %A Haichun Yang %A Yuankai Huo %A Mert R. Sabuncu %B Proceedings of The 8th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Tolga Tasdizen %E Shireen Elhabian %E Ronald Summers %E Chen Chen %E Lisa Koch %E Yan Zhuang %F pmlr-v301-deng26a %I PMLR %P 293--309 %U https://proceedings.mlr.press/v301/deng26a.html %V 301 %X Multi-class cell segmentation in high-resolution gigapixel whole slide images (WSIs) is crucial for various clinical applications. However, training such models typically requires labor-intensive, pixel-wise annotations by domain experts. Recent efforts have democratized this process by involving lay annotators without medical expertise. However, conventional non-corrective approaches struggle to handle annotation noise adaptively because they lack mechanisms to mitigate false positives (FP) and false negatives (FN) at both the image-feature and pixel levels. In this paper, we propose a consensus-aware self-corrective learning that leverages the Consensus Matrix to guide its learning process. The Consensus Matrix defines regions where both the AI and annotators agree on cell and non-cell annotations, which are prioritized with stronger supervision. Conversely, areas of disagreement are adaptively weighted based on their feature similarity to high-confidence consensus regions, with more similar regions receiving greater attention. Additionally, contrastive learning is employed to separate features of noisy regions from those of reliable consensus regions by maximizing their dissimilarity. This paradigm enables the model to iteratively refine noisy labels, enhancing its robustness. Validated on one real-world lay-annotated cell dataset and two reasoning-guided simulated noisy datasets, our method demonstrates improved segmentation performance, effectively correcting FP and FN errors and showcasing its potential for training robust models on noisy datasets. The official implementation and cell annotations are publicly available at https://github.com/ddrrnn123/CASC-AI.
APA
Deng, R., Yang, Y., Pisapia, D.J., Liechty, B.L., Zhu, J., Xiong, J., Guo, J., Lu, Z., Wang, J., Yao, X., Yu, R., Zhang, R., Rudravaram, G., Yin, M., Sarder, P., Yang, H., Huo, Y. & Sabuncu, M.R.. (2026). CASC-AI: Consensus-aware Self-corrective Learning for Cell Segmentation with Noisy Labels. Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 301:293-309 Available from https://proceedings.mlr.press/v301/deng26a.html.

Related Material