Region Constraint-Based Recognition and Localization Network of Multi-Instrument for Robotic Laparoscope Holders
IEEE Transactions on Medical Robotics and Bionics · 2026-08 · 8 (3) · 1133-1142
DOI: 10.1109/tmrb.2026.3712797 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Medical Robotics and Bionics, 2026-08. Verfasst von Jiayi Zhang und 7 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Taishan Industry Leading Talents, Suzhou Key Laboratory.
Abstract (Originalsprache)
In laparoscopic surgery, robotic laparoscope holders can autonomously guide the laparoscope, replacing human camera assistants to provide a stable surgical field of view. However, real-time and accurate recognition and localization of surgical instruments are critical for autonomous laparoscope guidance, yet they are significantly hindered by challenges such as blood occlusion and specular reflection. To address this, our study proposes a region constraint-based multi-instrument recognition and localization network (RC-MRLNet), which employs a multi-task learning strategy. Our network framework integrates a multi-instance segmentation network (MSNet) and a saliency map regression network (SRNet) to simultaneously achieve multi-instrument recognition and localization. Additionally, a semantic segmentation network (SSNet) refines performance by extracting shaft and manipulator regions, while a novel two-stage multi-target cooperative optimization strategy enhances training coordination. Evaluated on public datasets from different laparoscopic surgeries, RC-MRLNet achieves real-time inference at 65 frames per second (FPS) with superior accuracy: a dice similarity coefficient (DSC) of 0.967 and saliency map similarity (SMS) of 0.365, outperforming existing methods by 1.5% and 20.6%, respectively. Cross-domain testing further demonstrates its strong generalization capability. This study presents an effective solution for real-time and accurate recognition and localization of surgical instruments, showcasing significant potential to advance robotic laparoscope assistance in minimally invasive surgery.
Förderung
- National Natural Science Foundation of China
- Taishan Industry Leading Talents
- Suzhou Key Laboratory
Zitieren
APA
Zhang, J., Zhang, R., Shi, W., Zhu, Y., To, S., Wang, W., Chen, Y., & Gao, X. (2026). Region Constraint-Based Recognition and Localization Network of Multi-Instrument for Robotic Laparoscope Holders. IEEE Transactions on Medical Robotics and Bionics, 8(3), 1133-1142. https://doi.org/10.1109/tmrb.2026.3712797
Vancouver
Zhang J, Zhang R, Shi W, Zhu Y, To S, Wang W, et al. Region Constraint-Based Recognition and Localization Network of Multi-Instrument for Robotic Laparoscope Holders. IEEE Transactions on Medical Robotics and Bionics. 2026;8(3):1133-1142. doi:10.1109/tmrb.2026.3712797
BibTeX
@article{Zhang2026,
author = {Zhang, Jiayi and Zhang, Rui and Shi, Wei and Zhu, Yechen and To, Suet and Wang, Wenkui and Chen, Yu and Gao, Xin},
title = {Region Constraint-Based Recognition and Localization Network of Multi-Instrument for Robotic Laparoscope Holders},
journal = {IEEE Transactions on Medical Robotics and Bionics},
year = {2026},
volume = {8},
number = {3},
pages = {1133--1142},
doi = {10.1109/tmrb.2026.3712797}
}
EUVIMED