EUVIMEDEUVIMED entdecken
Crossref

Stitching, Fine-Tuning, and Re-Training: A SAM-Enabled Framework for Semi-Supervised 3D Medical Image Segmentation

Shumeng Li, Lei Qi, Qian Yu, Jing Huo, Yinghuan Shi, Yang Gao

IEEE Transactions on Medical Imaging · 2025-10 · 44 (10) · 3909-3923

DOI: 10.1109/tmi.2025.3532084 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Medical Imaging, 2025-10. Verfasst von Shumeng Li und 5 weiteren Autor:innen. Hauptthemen laut MEDLINE: Imaging, Three-Dimensional, Supervised Machine Learning. Die Arbeit wurde bisher 26-mal zitiert (Crossref) und nennt 60 Literaturangaben. Gefördert von: NSFC Project, China Post-Doctoral Science Foundation, Fundamental Research Funds for the Central Universities.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Segment Anything Model (SAM) fine-tuning has shown remarkable performance in medical image segmentation in a fully supervised manner, but requires precise annotations. To reduce the annotation cost and maintain satisfactory performance, in this work, we leverage the capabilities of SAM for establishing semi-supervised medical image segmentation models. Rethinking the requirements of effectiveness, efficiency, and compatibility, we propose a three-stage framework, i.e., Stitching, Fine-tuning, and Re-training (SFR). The current fine-tuning approaches mostly involve 2D slice-wise fine-tuning that disregards the contextual information between adjacent slices. Our stitching strategy mitigates the mismatch between natural and 3D medical images. The stitched images are then used for fine-tuning SAM, providing robust initialization of pseudo-labels. Afterwards, we train a 3D semi-supervised segmentation model while maintaining the same parameter size as the conventional segmenter such as V-Net. Our SFR framework is plug-and-play, and easily compatible with various popular semi-supervised methods. We also develop an extended framework SFR+ with selective fine-tuning and re-training through confidence estimation. Extensive experiments validate that our SFR and SFR+ achieve significant improvements in both moderate annotation and scarce annotation across five datasets. In particular, SFR framework improves the Dice score of Mean Teacher from 29.68% to 74.40% with only one labeled data of LA dataset. The code is available at https://github.com/ShumengLI/SFR.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Imaging, Three-DimensionalSupervised Machine LearningHumansAlgorithmsMagnetic Resonance Imaging

Förderung

  • NSFC Project
  • China Post-Doctoral Science Foundation
  • Fundamental Research Funds for the Central Universities
  • State Key Laboratory Fund
  • Post-Doctoral Fellowship Programof China Postdoctoral Science Foundation
  • Jiangsu Funding Program for Excellent Post-Doctoral Talent
  • Jiangsu Science and Technology Major Project
  • Natural Science Foundation of Shandong Province

Zitieren

APA

Li, S., Qi, L., Yu, Q., Huo, J., Shi, Y., & Gao, Y. (2025). Stitching, Fine-Tuning, and Re-Training: A SAM-Enabled Framework for Semi-Supervised 3D Medical Image Segmentation. IEEE Transactions on Medical Imaging, 44(10), 3909-3923. https://doi.org/10.1109/tmi.2025.3532084

Vancouver

Li S, Qi L, Yu Q, Huo J, Shi Y, Gao Y. Stitching, Fine-Tuning, and Re-Training: A SAM-Enabled Framework for Semi-Supervised 3D Medical Image Segmentation. IEEE Transactions on Medical Imaging. 2025;44(10):3909-3923. doi:10.1109/tmi.2025.3532084

BibTeX
@article{Li2025,
  author = {Li, Shumeng and Qi, Lei and Yu, Qian and Huo, Jing and Shi, Yinghuan and Gao, Yang},
  title = {Stitching, Fine-Tuning, and Re-Training: A SAM-Enabled Framework for Semi-Supervised 3D Medical Image Segmentation},
  journal = {IEEE Transactions on Medical Imaging},
  year = {2025},
  volume = {44},
  number = {10},
  pages = {3909--3923},
  doi = {10.1109/tmi.2025.3532084}
}
Kostenlos starten →