SfMDiffusion: self-supervised monocular depth estimation in endoscopy based on diffusion models
International Journal of Computer Assisted Radiology and Surgery · 24.02.2025 · 20 (5) · 971-979
DOI: 10.1007/s11548-025-03333-0 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computer Assisted Radiology and Surgery, online erschienen am 24.02.2025. Verfasst von Yu Li und 7 weiteren Autor:innen. Beteiligte Einrichtung: The Institute of Technological Sciences, Wuhan University, Wuhan, China. Hauptthemen laut MEDLINE: Imaging, Three-Dimensional, Laparoscopy, Surgery, Computer-Assisted. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 30 Literaturangaben. Gefördert von: Key Technologies Research and Development Program of Anhui Province, Fundamental Research Funds for the Central Universities, Science and Technology Program of Hubei Province.
Abstract (Originalsprache)
PURPOSE: In laparoscopic surgery, accurate 3D reconstruction from endoscopic video is crucial for effective image-guided techniques. Current methods for monocular depth estimation (MDE) face challenges in complex surgical scenes, including limited training data, specular reflections, and varying illumination conditions. METHODS: We propose SfMDiffusion, a novel diffusion-based self-supervised framework for MDE. Our approach combines: (1) a denoising diffusion process guided by pseudo-ground-truth depth maps, (2) knowledge distillation from a pre-trained teacher model, and (3) discriminative priors to enhance estimation robustness. Our design enables accurate depth estimation without requiring ground-truth depth data during training. RESULTS: Experiments on the SCARED and Hamlyn datasets demonstrate that SfMDiffusion achieves superior performance: an Absolute relative error (Abs Rel) of 0.049, a Squared relative error (Sq Rel) of 0.366, and a Root Mean Square Error (RMSE) of 4.305 on SCARED dataset, and Abs Rel of 0.067, Sq Rel of 0.800, and RMSE of 7.465 on Hamlyn dataset. CONCLUSION: SfMDiffusion provides an innovative approach for 3D reconstruction in image-guided surgical techniques. Future work will focus on computational optimization and validation across diverse surgical scenarios. Our code is available at https://github.com/Skylanding/SfM-Diffusion .
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- The Institute of Technological Sciences, Wuhan University, Wuhan, China.
- Department of Networked Intelligence, Pengcheng Laboratory, Shenzhen, China.
- School of Computer Science, Hubei University of Technology, Wuhan, China.
- Department of Gynecology, Renmin Hospital of Wuhan University, Wuhan, China.
- The Institute of Technological Sciences, Wuhan University, Wuhan, China. liyemei@whu.edu.cn.
- Department of Networked Intelligence, Pengcheng Laboratory, Shenzhen, China. liyemei@whu.edu.cn.
Förderung
- Key Technologies Research and Development Program of Anhui Province
- Fundamental Research Funds for the Central Universities
- Science and Technology Program of Hubei Province
Zitieren
APA
Li, Y., Chang, D., Luo, D., Huang, J., Dong, L., Wang, D., Mei, L., & Lei, C. (2025). SfMDiffusion: self-supervised monocular depth estimation in endoscopy based on diffusion models. International Journal of Computer Assisted Radiology and Surgery, 20(5), 971-979. https://doi.org/10.1007/s11548-025-03333-0
Vancouver
Li Y, Chang D, Luo D, Huang J, Dong L, Wang D, et al. SfMDiffusion: self-supervised monocular depth estimation in endoscopy based on diffusion models. International Journal of Computer Assisted Radiology and Surgery. 2025;20(5):971-979. doi:10.1007/s11548-025-03333-0
BibTeX
@article{Li2025,
author = {Li, Yu and Chang, Da and Luo, Die and Huang, Jin and Dong, Lan and Wang, Du and Mei, Liye and Lei, Cheng},
title = {SfMDiffusion: self-supervised monocular depth estimation in endoscopy based on diffusion models},
journal = {International Journal of Computer Assisted Radiology and Surgery},
year = {2025},
volume = {20},
number = {5},
pages = {971--979},
doi = {10.1007/s11548-025-03333-0}
}
EUVIMED