Thyroid nodule segmentation in ultrasound images using transformer models with masked autoencoder pre-training
MAE pre-training notably reduces training time and helps themodel learn transferable features, yet overall accuracy remains constrained by limited data and nodule variability. Future work will focus on scaling up data, pre-training cross-attention layers, and exploring hybrid architectures to further boost segmentation performance.
Authors
- Chiaw Ling Chng (ORCID: https://orcid.org/0000-0003-2232-5298)
- Yi Xiang (ORCID: https://orcid.org/0000-0003-2118-4825)
- Jen Hong Tan (ORCID: https://orcid.org/0000-0002-7785-2987)
- Le Quan (ORCID: https://orcid.org/0009-0002-9787-2464)
Institutions
- SingHealth (SG)
- University of Southern Queensland (AU)
- Singapore General Hospital (SG)
Publication Details
- Journal
- Frontiers in Artificial Intelligence
- Published
- 2025-07-24
- DOI
- https://doi.org/10.3389/frai.2025.1618426
- Citations
- 2
- Primary Topic
- AI in cancer detection
- Type
- article
- Field-Weighted Citation Impact
- 4.96
Funders
- Duke-NUS Medical School
- Singapore General Hospital