Impact of differential privacy on breast ultrasound image classification performance using vision transformer
Our systematic evaluation demonstrates that DP can be integrated into breast ultrasound classification while maintaining quantifiable privacy guarantees. Selecting ε≥1 and δ=0.1 preserves diagnostic accuracy close to the non-private baseline and provides a practical reference for privacy-preserving clinical deployment.
Authors
- Yanting Wen (ORCID: https://orcid.org/0000-0002-1049-5110)
- Zhigang Luo (ORCID: https://orcid.org/0000-0002-7552-201X)
- Xiaofei Zhong (ORCID: https://orcid.org/0000-0002-9550-1769)
- Jiaqi Yang (ORCID: https://orcid.org/0000-0002-7038-0913)
Institutions
- Sichuan University (CN)
- West China Hospital of Sichuan University (CN)
Publication Details
- Journal
- Quantitative Imaging in Medicine and Surgery
- Published
- 2025-11-25
- DOI
- https://doi.org/10.21037/qims-2025-1064
- Primary Topic
- AI in cancer detection
- Type
- article
- Field-Weighted Citation Impact
- 0.00