Paced-curriculum distillation with prediction and label uncertainty for image segmentation
File(s)
Accepted version
Author(s)
Type
Journal Article
Abstract
PURPOSE: In curriculum learning, the idea is to train on easier samples first and gradually increase the difficulty, while in self-paced learning, a pacing function defines the speed to adapt the training progress. While both methods heavily rely on the ability to score the difficulty of data samples, an optimal scoring function is still under exploration. METHODOLOGY: Distillation is a knowledge transfer approach where a teacher network guides a student network by feeding a sequence of random samples. We argue that guiding student networks with an efficient curriculum strategy can improve model generalization and robustness. For this purpose, we design an uncertainty-based paced curriculum learning in self-distillation for medical image segmentation. We fuse the prediction uncertainty and annotation boundary uncertainty to develop a novel paced-curriculum distillation (P-CD). We utilize the teacher model to obtain prediction uncertainty and spatially varying label smoothing with Gaussian kernel to generate segmentation boundary uncertainty from the annotation. We also investigate the robustness of our method by applying various types and severity of image perturbation and corruption. RESULTS: The proposed technique is validated on two medical datasets of breast ultrasound image segmentation and robot-assisted surgical scene segmentation and achieved significantly better performance in terms of segmentation and robustness. CONCLUSION: P-CD improves the performance and obtains better generalization and robustness over the dataset shift. While curriculum learning requires extensive tuning of hyper-parameters for pacing function, the level of performance improvement suppresses this limitation.
Date Issued
2023-10
Date Acceptance
2023-01-31
Citation
International Journal of Computer Assisted Radiology and Surgery, 2023, 18 (10), pp.1875-1883
ISSN
1861-6410
Publisher
Springer
Start Page
1875
End Page
1883
Journal / Book Title
International Journal of Computer Assisted Radiology and Surgery
Volume
18
Issue
10
Copyright Statement
Copyright © 2023 Springer-Verlag. This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1007/s11548-023-02847-9
Identifier
https://www.ncbi.nlm.nih.gov/pubmed/36862365
Subjects
Boundary uncertainty
Curriculum learning
Distillation
Segmentation
Publication Status
Published
Coverage Spatial
Germany
Date Publish Online
2023-03-02