Published on: May 2026
LIVERSEGNET: A HYBRID CONTEXT-AWARE PERCEPTION FRAMEWORK FOR SAFETY-CRITICAL NAVIGATION IN LAPAROSCOPIC SURGERY
Akash Kumar Saket Bishnu Anshit Omveer Singh
Dr. Annapoorani S
Article Status
Available Documents
Abstract
— Laparoscopic surgery presents an adversarial environment for machine perception, characterized by dynamic occlusion, specular artifacts, and attenuated lighting. While deep learning models achieve high accuracy on curated datasets, their lack of transparency and propensity for semantic flicker in shadowed regions pose significant risks for clinical navigation. We propose LiverSegNet, a hybrid context-aware perception framework that transcends “pure- neural” limitations by fusing a dual-kernel neural perception layer with physically informed heuristics and deterministic safety guardrails. Our system addresses the “Black-Box” dilemma by grounding pattern-based neural intuition in the physical laws of optics and geometric constraints. In this work, we formally derive a custom Surgical Hybrid Loss— utilizing the Focal Tversky Index—to mathematically shift the system’s objective toward anatomical recall (safety) rather than generic precision. Furthermore, we implement a Multicolour Recovery (MAR) module that reclaims anatomy in deep shadows via hue-locked growth kernels. Experimental validation on the CholecSeg8K and CholecInstanceSeg datasets demonstrates an Intersection over Union (IoU) of 84.56% at a real-time latency of 32ms. Crucially, we introduce an Explainability (XAI) component via Heatmap Diagnostics, enabling real-time uncertainty quantification for the surgical team. This hybrid approach provides a redundant, transparent, and robust alternative to standard segmentation baselines, aligning with the safety-first requirements of high- stakes intraoperative environments.
How to Cite this Paper
Kumar, A., Bishnu, S. & Singh, A. O. (2026). Liversegnet: A Hybrid Context-Aware Perception Framework for Safety-Critical Navigation in Laparoscopic Surgery. International Journal of Creative and Open Research in Engineering and Management, <i>02</i>(05). https://doi.org/10.55041/ijcope.v2i4.944
Kumar, Akash, et al.. "Liversegnet: A Hybrid Context-Aware Perception Framework for Safety-Critical Navigation in Laparoscopic Surgery." International Journal of Creative and Open Research in Engineering and Management, vol. 02, no. 05, 2026, pp. . doi:https://doi.org/10.55041/ijcope.v2i4.944.
Kumar, Akash,Saket Bishnu, and Anshit Singh. "Liversegnet: A Hybrid Context-Aware Perception Framework for Safety-Critical Navigation in Laparoscopic Surgery." International Journal of Creative and Open Research in Engineering and Management 02, no. 05 (2026). https://doi.org/https://doi.org/10.55041/ijcope.v2i4.944.
Search & Index
References
Conclusion- LiverSegNet proves that surgical AI must transcend pure probability to achieve clinical trust. By fusing neural intuition with physical heuristics and deterministic rules, we have created a perception framework that is both robust and transparent. Our work provides a blueprint for “Human-in- the-Loop” surgical systems that enhance, rather than replace, clinical decision-making. By prioritizing anatomical recall and safety-first loss functions, we move one step closer to a future where surgical complications reach a statistical minimum.
Acknowledgments
- We thank the clinical advisors and the surgical data science community for providing the CholecSeg8K and CholecInstanceSeg datasets, which made this research possible.
References
- B. Hanna, A. Shimi, and M. Cuschieri, “Task performance in endoscopic surgery is influenced by location of the image display,” Annals of Surgery, vol. 227, no. 4, pp. 481–484, 1998.
- R. Lanfranco, A. E. Castellanos, J. P. Desai, and W. C. Meyers, “Robotic surgery: a current perspective,” Annals of Surgery, vol. 239, no. 1, pp. 14–21, 2004.
- Twinanda et al., “EndoNet: A deep architecture for recognition tasks on laparoscopic videos,” IEEE Trans. Med. Imaging, vol. 36, no. 1, pp. 86–97, 2017.
- Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional networks for biomedical image segmentation,” in MICCAI, 2015,
- 234–241.
- Twinanda et al., “EndoNet dataset (Cholec80),” arXiv preprint arXiv:1602.03012, 2016.
- Rivoir et al., “CholecSeg8k: A semantic segmentation dataset for laparoscopic surgery,” Scientific Data, vol. 8, no. 1, 2021.
- Nwoye et al., “CholecSeg8k and CholecInstanceSeg datasets,”arXiv:2109.04242, 2021.
- -C. Chen et al., “Encoder-decoder with atrous separable convolution for semantic image segmentation,” in ECCV, 2018.
- Wang et al., “Deep high-resolution representation learning for visual recognition,” IEEE TPAMI, vol. 43, no. 10, pp. 3349–3364, 2021.
- Chen et al., “TransUNet: Transformers make strong encoders for medical image segmentation,” arXiv:2102.04306, 2021.
- Raissi, P. Perdikaris, and G. E. Karniadakis, “Physics-informed neural networks: A deep learning framework for solving forward and inverse problems,” J. Comput. Phys., vol. 378, pp. 686–707, 2019.
- He et al., “Deep residual learning for image recognition,” in CVPR, 2016.Simonyan and A. Zisserman, “Very deep convolutional networks
- for large-scale image recognition,” arXiv:1409.1556, 2014.S. Salehi et al., “Tversky loss function for image segmentation
- using 3D fully convolutional deep networks,” in MLMI, 2017.-Y. Lin et al., “Focal loss for dense object detection,” in ICCV, 2017.
- C. Gonzalez and R. E. Woods, Digital Image Processing, 4th ed., Pearson, 2018.
- Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning, Springer, 2009.
- Everingham et al., “The Pascal Visual Object Classes (VOC) challenge,” IJCV, vol. 88, no. 2, pp. 303–338, 2010.
- R. Selvaraju et al., “Grad-CAM: Visual explanations from deep networks via gradient-based localization,” in ICCV, 2017.
- J. Topol, “High-performance medicine: the convergence of human and artificial intelligence,” Nature Medicine, vol. 25, pp. 44– 56, 2019.
Ethical Compliance & Review Process
- •All submissions are screened under plagiarism detection.
- •Review follows editorial policy.
- •Authors retain copyright.
- •Peer Review Type: Double-Blind Peer Review
- •Published on: May 07 2026
This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. You are free to share and adapt this work for non-commercial purposes with proper attribution.