Unsupervised domain adaptation for medical imaging segmentation with self-ensembling

By Christian S. Perone, Pedro L. Ballester, Rodrigo C. Barros, Julien Cohen-Adad. 2018

249 citations recorded by the source corpus.

Recent advances in deep learning methods have redefined the state-of-the-art for many medical imaging applications, surpassing previous approaches and sometimes even competing with human judgment in several tasks. Those models, however, when trained to reduce the empirical risk on a single domain, fail to generalize when applied to other domains, a very common scenario in medical imaging due to the variability of images and anatomical structures, even across the same imaging modality. In this work, we extend the method of unsupervised domain adaptation using self-ensembling for the semantic segmentation task and explore multiple facets of the method on a small and realistic publicly-available magnetic resonance (MRI) dataset. Through an extensive evaluation, we show that self-ensembling can indeed improve the generalization of the models even when using a small amount of unlabeled data.

Subfields: Artificial Intelligence, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. P. Heng
2. Qi-Yan Dou
3. Hao Chen
4. Jing Qin
5. Lequan Yu
6. Diane Joyce Cook
7. Garrett Wilson
8. Cheng Chen

In Radiology, Nuclear Medicine and Imaging

1. P. Heng
2. Qi-Yan Dou
3. Hao Chen
4. Lequan Yu
5. Jing Qin
6. Diane Joyce Cook
7. Garrett Wilson
8. Cheng Chen

Citations in the corpus

• In Search of Lost Domain Generalization (2020, neutral): “As another example, systems trained on medical data collected in one hospital do not generalize to other health centers [Castro et al., 2019, AlBadawy et al., 2018, Perone et al., 2019, Heaven, 2020].”
• Explainable, trustworthy, and ethical machine learning for healthcare: A survey (2021, background): “2) Robustness to Distributional Shifts and Data Imperfec-tions: Data distribution shifts (which refers to the divergence of training and testing data) is yet another major challenge that hinders the practical deployment of ML/DL models in realistic clinical settings [79].”
• Self-Ensembling With GAN-Based Data Augmentation for Domain Adaptation in Semantic Segmentation (2019, methodology): “In [35], Perone et al . address medical imaging segmentation tasks by applying the self-ensembling method akin to the previous method.”
• Domain Adaptation for Medical Image Analysis: A Survey (2021, unknown): “Through minimizing the triplet loss, their model can reduce the domain shift while keeping discrimination on different diseases.”
• Unsupervised Bidirectional Cross-Modality Adaptation via Deeply Synergistic Image and Feature Alignment for Medical Image Segmentation (2020, background): “Even when using the same MRI sequence, the data distribution can vary in datasets acquired at different centers [5], [29] or different time [4].”
• A Survey of Unsupervised Deep Domain Adaptation (2018, methodology): “a simulator [133], in a medical context for chest X-ray segmentation [32], 3D CT scans to Xray segmentation [277], MRI to CT scan segmentation [33], and MRI segmentation [185], in low resource situations (where there are very few target data points) [100], in situations with different label sets for”
• Deep Unsupervised Domain Adaptation: A Review of Recent Advances and Perspectives (2022, methodology): “Perone et al. (2019) use a self-ensembling technique in semantic image segmentation, demonstrating that it can improve model generalization.”
• A Noise-Robust Framework for Automatic Segmentation of COVID-19 Pneumonia Lesions From CT Images (2020, methodology): “Differently from previous self-ensembling methods for semi-supervised learning [20], [21] and domain adaptation [22], [23], we propose two adaptive mechanisms to better deal with noisy labels: adaptive teacher that suppresses the contribution of the student to EMA when the latter has a large…”
• Uncertainty-aware multi-view co-training for semi-supervised medical image segmentation and domain adaptation (2020, background): “In the field of medical image analysis, [13, 29, 45] have investigated this topic with existing approaches, i.”
• Secure and Robust Machine Learning for Healthcare: A Survey (2020, methodology): “8) Distribution Shifts: The problem of data distribution shifts is yet another major challenge and perhaps one of the most challenging problems to solve [136].”