Abstract
Insufficient data is always a big challenge for medical imaging that is limited by the expensive labeling cost, time-consuming and intensive labor. Active learning aims to reduce the annotation effort by training a model on actively selected samples, most of them adopt uncertainty measures as instance selection criteria. However, uncertainty strategies underperform in most active learning studies. In addition, inaccurate selections worse than random sampling in initial stage referred to as “cold start” problem is also a huge challenge for active learning. Domain adaptation aims at alleviating the cold start problem and also reducing the annotation effort by adapting the model from a pre-trained model trained on another domain. Our work focuses on whether active learning can benefit from domain adaptation and the performance of uncertainty strategy compared to random selection. We studied 3D hippocampus images segmentation based on 3D UX-Net and four MRI datasets Hammers, HarP, LPBA40, and OASIS. Our experiments reveal that active learning with domain adaptation is more efficient and robust than without domain adaptation at a low labeling budget. The performance gap between them diminishes as we approach to that half of the dataset is labeled. In addition, entropy sampling also converges faster than random sampling, with slightly better performance.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bengar, J.Z., van de Weijer, J., Twardowski, B., Raducanu, B.: Reducing label effort: self-supervised meets active learning (2021)
Boccardi, M., et al.: Training labels for hippocampal segmentation based on the EADC-ADNI harmonized hippocampal protocol. Alzheimer’s Dementia 11(2), 175–183 (2015). https://doi.org/10.1016/j.jalz.2014.12.002
Carmo, D., Silva, B., Yasuda, C., Rittner, L., Lotufo, R.: Hippocampus segmentation on epilepsy and Alzheimer’s disease studies with multiple convolutional neural networks. Heliyon 7(2), e06226 (2021). https://doi.org/10.1016/j.heliyon.2021.e06226
Casanova, A., Pinheiro, P.O., Rostamzadeh, N., Pal, C.J.: Reinforced active learning for image segmentation (2020)
Chen, L.C., Papandreou, G., Kokkinos, I., Murphy, K., Yuille, A.L.: DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. IEEE Trans. Pattern Anal. Mach. Intell. 40(4), 834–848 (2018). https://doi.org/10.1109/TPAMI.2017.2699184
Chen, L., et al.: Making your first choice: to address cold start problem in medical active learning. In: Medical Imaging with Deep Learning (2023). https://openreview.net/forum?id=5iSBMWm3ln
Chen, L., Fu, Y., You, S., Liu, H.: Hybrid supervised instance segmentation by learning label noise suppression. Neurocomputing 496, 131–146 (2022). https://doi.org/10.1016/j.neucom.2022.05.026
Dice, L.R.: Measures of the amount of ecologic association between species. Ecology 26(3), 297–302 (1945). https://www.jstor.org/stable/1932409
Dubois, B., Hampel, H., Feldman, H.H., Carrillo, M.C., Cummings, J., Jack, C.R., Jr.: Preclinical Alzheimer’s disease: definition, natural history, and diagnostic criteria. Alzheimers Dement. 12(3), 292–323 (2016)
Faillenot, I., Heckemann, R.A., Frot, M., Hammers, A.: Macroanatomy and 3D probabilistic atlas of the human insula. Neuroimage 150, 88–98 (2017). https://doi.org/10.1016/j.neuroimage.2017.01.073
Farahani, A., Voghoei, S., Rasheed, K., Arabnia, H.R.: A brief review of domain adaptation (2020)
Isensee, F., Jaeger, P.F., Kohl, S.A.A., Petersen, J., Maier-Hein, K.H.: nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation. Nat. Methods 18(2), 203–211 (2021)
Ji, Y., et al.: AMOS: a large-scale abdominal multi-organ benchmark for versatile medical image segmentation. arXiv preprint arXiv:2206.08023 (2022)
Konyushkova, K., Sznitman, R., Fua, P.: Learning active learning from data (2017)
Landman, B.A., Warfield, S.K.: MICCAI 2012: grand challenge and workshop on multi-atlas labeling. In: International Conference on MICCAI (2012)
Lee, H.H., Bao, S., Huo, Y., Landman, B.A.: 3D UX-Net: a large kernel volumetric convnet modernizing hierarchical transformer for medical image segmentation (2022). https://doi.org/10.48550/ARXIV.2209.15076
Loshchilov, I., Hutter, F.: Decoupled weight decay regularization (2019)
Ma, J., et al.: Fast and low-GPU-memory abdomen CT organ segmentation: the flare challenge. Med. Image Anal. 82, 102616 (2022)
Makili, L.E., Sánchez, J.A.V., Dormido-Canto, S.: Active learning using conformal predictors: application to image classification. Fusion Sci. Technol. 62(2), 347–355 (2012). https://doi.org/10.13182/FST12-A14626
Martin, J., Radzyner, H., Leonard, M.: Neuroanatomy: Text and Atlas. Ovid ebook collection, McGraw-Hill Companies, Incorporated (2003). https://books.google.dk/books?id=OUC4igr3O4sC
Mehdipour Ghazi, M., Nielsen, M.: FAST-AID brain: fast and accurate segmentation tool using artificial intelligence developed for brain. arXiv preprint (2022). https://doi.org/10.48550/ARXIV.2208.14360
Nath, V., Yang, D., Landman, B.A., Xu, D., Roth, H.R.: Diminishing uncertainty within the training pool: active learning for medical image segmentation. IEEE Trans. Med. Imaging 40(10), 2534–2547 (2021). https://doi.org/10.1109/tmi.2020.3048055
Owen, A.B.: Monte Carlo theory, methods and examples (2013)
Paszke, A., Gross, S., Massa, F., Bai, J., Chintala, S.: PyTorch: an imperative style, high-performance deep learning library. In: Advances in Neural Information Processing Systems, vol. 32, pp. 8024–8035. Curran Associates, Inc. (2019). https://papers.neurips.cc/paper/9015-pytorch-an-imperative-style-high-performance-deep-learning-library.pdf
Patel, K., Kovalyov, A., Panahi, I.: UX-Net: filter-and-process-based improved u-net for real-time time-domain audio separation (2022). https://doi.org/10.48550/ARXIV.2210.15822
Payette, K., de Dumast, P., Kebiri, H., Ezhov, I., Bach Cuadra, M., Jakab, A.: An automatic multi-tissue human fetal brain segmentation benchmark using the fetal tissue annotation dataset. Sci. Data 8(1), 167 (2021). https://doi.org/10.1038/s41597-021-00946-3
Postel, C., et al.: Variations in response to trauma and hippocampal subfield changes. Neurobiol. Stress 15, 100346 (2021). https://doi.org/10.1016/j.ynstr.2021.100346
Raj, A., Bach, F.: Convergence of uncertainty sampling for active learning (2021). https://doi.org/10.48550/ARXIV.2110.15784
Ronneberger, O., Fischer, P., Brox, T.: U-Net: convolutional networks for biomedical image segmentation (2015). https://doi.org/10.48550/ARXIV.1505.04597
Scheffer, T., Decomain, C., Wrobel, S.: Active hidden Markov models for information extraction. In: Hoffmann, F., Hand, D.J., Adams, N., Fisher, D., Guimaraes, G. (eds.) IDA 2001. LNCS, vol. 2189, pp. 309–318. Springer, Heidelberg (2001). https://doi.org/10.1007/3-540-44816-0_31
Shattuck, D., et al.: Construction of a 3D probabilistic atlas of human cortical structures. NeuroImage 39, 1064–1080 (2008). https://doi.org/10.1016/j.neuroimage.2007.09.031
Shelhamer, E., Long, J., Darrell, T.: Fully convolutional networks for semantic segmentation (2016). https://doi.org/10.48550/ARXIV.1605.06211
Sugiyama, M., Kawanabe, M.: Active learning with model selection, pp. 215–224 (2012)
Wu, D.: Pool-based sequential active learning for regression (2018)
Wu, T.H., et al.: D2ADA: dynamic density-aware active domain adaptation for semantic segmentation (2022)
Xie, B., Yuan, L., Li, S., Liu, C.H., Cheng, X., Wang, G.: Active learning for domain adaptation: an energy-based approach. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 36, no. 8 (2022)
Acknowledgements
This project has received funding from Pioneer Centre for AI, Danish National Research Foundation, grant number P1.
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
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Wu, J., Kang, Z., Llambias, S.N., Ghazi, M.M., Nielsen, M. (2023). Active Transfer Learning for 3D Hippocampus Segmentation. In: Xue, Z., et al. Medical Image Learning with Limited and Noisy Data. MILLanD 2023. Lecture Notes in Computer Science, vol 14307. Springer, Cham. https://doi.org/10.1007/978-3-031-44917-8_22
Download citation
DOI: https://doi.org/10.1007/978-3-031-44917-8_22
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-47196-4
Online ISBN: 978-3-031-44917-8
eBook Packages: Computer ScienceComputer Science (R0)