Abstract
Supervised training with deep learning has exhibited impressive performance in numerous medical image domains. However, previous successes rely on the availability of well-labeled data. In practice, it is a great challenge to obtain a large high-quality labeled dataset, especially for the medical image segmentation task, which generally needs pixel-wise labels, and the inaccurate label (noisy label) may significantly degrade the segmentation performance. In this paper, we propose a novel Meta Pixel Loss Correction (MPLC) based on a simple meta guided network for the medical segmentation that is robust to noisy labels. The core idea is to estimate a pixel transition confidence map by meta guided network to take full advantage of noisy labels for pixel-wise loss correction. To achieve this, we introduce a small size of meta dataset with the meta-learning method to train the whole model and help the meta guided network automatically learn the pixel transition confidence map in an alternative training manner. Experiments have been conducted on three medical image datasets, and the results demonstrate that our method is able to achieve superior segmentation with noisy labels compared to the existing state-of-the-art approaches.
Keywords
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Armato, S.G., et al.: The lung image database consortium (LIDC) and image database resource initiative (IDRI): a completed reference database of lung nodules on CT scans. Acad. Radiol. 14(12), 1455–1463 (2007)
Audelan, B., Delingette, H.: Unsupervised quality control of image segmentation based on Bayesian learning. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11765, pp. 21–29. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32245-8_3
Baumgartner, F., et al.: PHiSeg: capturing uncertainty in medical image segmentation. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11765, pp. 119–127. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32245-8_14
Han, X.: Automatic liver lesion segmentation using a deep convolutional neural network method (2017)
Hu, J., Shen, L., Albanie, S., Sun, G., Wu, E.: Squeeze-and-excitation networks. IEEE Trans. Pattern Anal. Mach. Intell. 42(8), 2011–2023 (2020). https://doi.org/10.1109/TPAMI.2019.2913372
Huang, C., Han, H., Yao, Q., Zhu, S., Zhou, S.K.: 3D U\(^2\)-net: a 3D universal u-net for multi-domain medical image segmentation. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11765, pp. 291–299. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32245-8_33
Karimi, D., Dou, H., Warfield, S.K., Gholipour, A.: Deep learning with noisy labels: Exploring techniques and remedies in medical image analysis. Med. Image Anal. 65, 101759 (2020)
Kervadec, H., Dolz, J., Granger, É., Ben Ayed, I.: Curriculum semi-supervised segmentation. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11765, pp. 568–576. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32245-8_63
Kohl, S.A., et al.: A probabilistic u-net for segmentation of ambiguous images. arXiv preprint arXiv:1806.05034 (2018)
Menze, B.H., Jakab, A., Bauer, S., et al.: The multimodal brain tumor image segmentation benchmark (brats). IEEE Trans. Med. Imaging 34(10), 1993–2024 (2015). https://doi.org/10.1109/TMI.2014.2377694
Mirikharaji, Z., Yan, Y., Hamarneh, G.: Learning to segment skin lesions from noisy annotations. In: Wang, Q., et al. (eds.) DART/MIL3ID -2019. LNCS, vol. 11795, pp. 207–215. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-33391-1_24
Mondal, A.K., Dolz, J., Desrosiers, C.: Few-shot 3d multi-modal medical image segmentation using generative adversarial learning. arXiv preprint arXiv:1810.12241 (2018)
Ren, M., Zeng, W., Yang, B., Urtasun, R.: Learning to reweight examples for robust deep learning. In: International Conference on Machine Learning, pp. 4334–4343. PMLR (2018)
Ronneberger, O., Fischer, P., Brox, T.: U-net: convolutional networks for biomedical image segmentation. In: Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F. (eds.) MICCAI 2015. LNCS, vol. 9351, pp. 234–241. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-24574-4_28
Shu, J., et al.: Meta-weight-net: learning an explicit mapping for sample weighting. arXiv preprint arXiv:1902.07379 (2019)
Song, H., Kim, M., Park, D., Shin, Y., Lee, J.G.: Learning from noisy labels with deep neural networks: a survey. IEEE Transactions on Neural Networks and Learning Systems (2022)
van Tulder, G.: Package elsticdeform. http://github.com/gvtulder/elasticdeform/. Accessed 4 Dec 2018
Wang, J., Zhou, S., Fang, C., Wang, L., Wang, J.: Meta corrupted pixels mining for medical image segmentation. In: Martel, A.L., et al. (eds.) MICCAI 2020. LNCS, vol. 12261, pp. 335–345. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-59710-8_33
Yu, L., Wang, S., Li, X., Fu, C.-W., Heng, P.-A.: Uncertainty-aware self-ensembling model for semi-supervised 3D left atrium segmentation. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11765, pp. 605–613. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32245-8_67
Zhu, H., Shi, J., Wu, J.: Pick-and-learn: automatic quality evaluation for noisy-labeled image segmentation. In: Shen, D., et al. (eds.) MICCAI 2019. LNCS, vol. 11769, pp. 576–584. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-32226-7_64
Acknowledgment
This work was supported by by the National Natural Science Foundation of China under Grant 61790562 and Grant 61773312.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
1 Electronic supplementary material
Below is the link to the electronic supplementary material.
Rights and permissions
Copyright information
© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Cai, Z., Xin, J., Shi, P., Zhou, S., Wu, J., Zheng, N. (2022). Meta Pixel Loss Correction for Medical Image Segmentation with Noisy Labels. In: Zamzmi, G., Antani, S., Bagci, U., Linguraru, M.G., Rajaraman, S., Xue, Z. (eds) Medical Image Learning with Limited and Noisy Data. MILLanD 2022. Lecture Notes in Computer Science, vol 13559. Springer, Cham. https://doi.org/10.1007/978-3-031-16760-7_4
Download citation
DOI: https://doi.org/10.1007/978-3-031-16760-7_4
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-16759-1
Online ISBN: 978-3-031-16760-7
eBook Packages: Computer ScienceComputer Science (R0)