Warm-started reinforcement learning for iterative 3D/2D liver registration
International Journal of Computer Assisted Radiology and Surgery · 20.05.2026 · 21 (6) · 1331-1339
DOI: 10.1007/s11548-026-03653-9 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computer Assisted Radiology and Surgery, online erschienen am 20.05.2026. Verfasst von Hanyuan Zhang und 7 weiteren Autor:innen. Beteiligte Einrichtung: UCL Hawkes Institute, University College London, London, UK. hanyuan.zhang.23@ucl.ac.uk. Hauptthemen laut MEDLINE: Reinforcement Machine Learning, Laparoscopy, Imaging, Three-Dimensional, Tomography, X-Ray Computed. Die Arbeit nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Wellcome Trust, National Institute for Health and Care Research, UK Research and Innovation.
Abstract (Originalsprache)
Purpose
Registration between preoperative CT and intraoperative laparoscopic video plays a crucial role in augmented reality (AR) guidance for minimally invasive surgery. Learning-based methods have recently achieved registration errors comparable to optimization-based approaches while offering faster inference. However, many supervised methods produce coarse alignments that rely on additional optimization-based refinement, thereby increasing inference time.
Methods
We present a discrete-action reinforcement learning (RL) framework that formulates CT-to-video registration as a sequential decision-making process. A shared feature encoder, warm-started from a supervised pose estimation network to provide stable geometric features and faster convergence, extracts representations from CT renderings and laparoscopic frames, while an RL policy head learns to choose rigid transformations along six degrees of freedom and to decide when to stop the iteration.
Results
Experiments on a public laparoscopic dataset demonstrated that our method achieved an average target registration error (TRE) of 15.70 ± 8.18 mm, comparable to supervised approaches with optimization, while achieving faster convergence.
Conclusion
The proposed RL-based formulation enables automated, efficient iterative registration without manually tuned step sizes or stopping criteria. This discrete framework provides a practical foundation for future continuous-action and deformable registration models in surgical AR applications.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- UCL Hawkes Institute, University College London, London, UK. hanyuan.zhang.23@ucl.ac.uk.
- UCL Hawkes Institute, University College London, London, UK.
- Medtronic plc., London, UK.
Förderung
- Wellcome Trust
- National Institute for Health and Care Research
- UK Research and Innovation
- Engineering and Physical Sciences Research Council
Zitieren
APA
Zhang, H., He, L., Cheng, Z., Kadkhodamohammadi, A., Stoyanov, D., Davidson, B. R., Mazomenos, E. B., & Clarkson, M. J. (2026). Warm-started reinforcement learning for iterative 3D/2D liver registration. International Journal of Computer Assisted Radiology and Surgery, 21(6), 1331-1339. https://doi.org/10.1007/s11548-026-03653-9
Vancouver
Zhang H, He L, Cheng Z, Kadkhodamohammadi A, Stoyanov D, Davidson BR, et al. Warm-started reinforcement learning for iterative 3D/2D liver registration. International Journal of Computer Assisted Radiology and Surgery. 2026;21(6):1331-1339. doi:10.1007/s11548-026-03653-9
BibTeX
@article{Zhang2026,
author = {Zhang, Hanyuan and He, Lucas and Cheng, Zijie and Kadkhodamohammadi, Abdolrahim and Stoyanov, Danail and Davidson, Brian R and Mazomenos, Evangelos B and Clarkson, Matthew J},
title = {Warm-started reinforcement learning for iterative 3D/2D liver registration},
journal = {International Journal of Computer Assisted Radiology and Surgery},
year = {2026},
volume = {21},
number = {6},
pages = {1331--1339},
doi = {10.1007/s11548-026-03653-9}
}
EUVIMED