Combining curriculum learning and weakly supervised attention for enhanced thyroid nodule assessment in ultrasound imaging
1
Issued Date
2025-09-01
Resource Type
ISSN
22234292
eISSN
22234306
Scopus ID
2-s2.0-105014268395
Journal Title
Quantitative Imaging in Medicine and Surgery
Volume
15
Issue
9
Start Page
8579
End Page
8593
Rights Holder(s)
SCOPUS
Bibliographic Citation
Quantitative Imaging in Medicine and Surgery Vol.15 No.9 (2025) , 8579-8593
Suggested Citation
Keatmanee C., Songsaeng D., Klabwong S., Nakaguro Y., Kunapinun A., Ekpanyapong M., Dailey M.N. Combining curriculum learning and weakly supervised attention for enhanced thyroid nodule assessment in ultrasound imaging. Quantitative Imaging in Medicine and Surgery Vol.15 No.9 (2025) , 8579-8593. 8593. doi:10.21037/qims-24-2431 Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/111925
Title
Combining curriculum learning and weakly supervised attention for enhanced thyroid nodule assessment in ultrasound imaging
Corresponding Author(s)
Other Contributor(s)
Abstract
Background: The accurate assessment of thyroid nodules, which are increasingly common with age and lifestyle factors, is essential for early malignancy detection. Ultrasound imaging, the primary diagnostic tool for this purpose, holds promise when paired with deep learning. However, challenges persist with small datasets, where conventional data augmentation can introduce noise and obscure essential diagnostic features. To address dataset imbalance and enhance model generalization, this study integrates curriculum learning with a weakly supervised attention network to improve diagnostic accuracy for thyroid nodule classification. Methods: This study integrates curriculum learning with attention-guided data augmentation to improve deep learning model performance in classifying thyroid nodules. Using verified datasets from Siriraj Hospital, the model was trained progressively, beginning with simpler images and gradually incorporating more complex cases. This structured learning approach is designed to enhance the model’s diagnostic accuracy by refining its ability to distinguish benign from malignant nodules. Results: Among the curriculum learning schemes tested, schematic IV achieved the best results, with a precision of 100% for benign and 70% for malignant nodules, a recall of 82% for benign and 100% for malignant, and F1-scores of 90% and 83%, respectively. This structured approach improved the model’s diagnostic sensitivity and robustness. Conclusions: These findings suggest that automated thyroid nodule assessment, supported by curriculum learning, has the potential to complement radiologists in clinical practice, enhancing diagnostic accuracy and aiding in more reliable malignancy detection.