ESD-VesNet: uncertainty-aware vessel segmentation network for endoscopic submucosal dissection with hard negative mining - Report - MDSpire

ESD-VesNet: uncertainty-aware vessel segmentation network for endoscopic submucosal dissection with hard negative mining

  • By

  • Mengya Xu

  • Ming Chen

  • Zhen Li

  • Chaoyang Lyu

  • An Wang

  • Rulin Zhou

  • Chuanhao Zhao

  • Jiaxun Xiang

  • Tsz Chun Wong

  • Hossein Farahnaki

  • Sobhan Zamani Kiasari

  • Tong Wu

  • Zimeng Su

  • Yile Zeng

  • Ruijing Wen

  • Xiaohan Shang

  • Yi Mu

  • Kezhen Lin

  • Yidong Zhang

  • Hongliang Ren

  • July 3, 2026

  • 0 min

Share

Clinical Report: Uncertainty-Driven Vessel Segmentation Network for ESD

Background

Endoscopic submucosal dissection (ESD) is a critical technique for treating early-stage gastrointestinal cancers, but it carries risks of intraoperative bleeding due to the complex vascular structures involved. Accurate vessel detection is essential to prevent complications during dissection, as human error can lead to significant adverse events.

Data Highlights

The ESD-Vessel dataset includes 44 high-definition surgical procedures with 2401 annotated vessel frames and 708 hard negative frames.

Key Findings

  • The ESD-VesNet framework utilizes the Segment Anything Model 3 (SAM3) for vessel segmentation in ESD.
  • Incorporation of hard negative mining techniques helps suppress false positives during training.
  • The framework achieves high vessel detection rates while maintaining low false-positive rates.
  • Approximately 20% of intraoperative adverse events are attributed to human visual perception errors.
  • Endoscopists achieve an average vessel detection rate of only about 56.4% without computational assistance.

Clinical Implications

The ESD-VesNet framework could provide endoscopists with enhanced tools for real-time vessel detection, potentially reducing the risk of intraoperative bleeding. Improved accuracy in vessel segmentation may lead to safer ESD procedures and better patient outcomes.

Conclusion

The introduction of ESD-VesNet represents a significant step forward in addressing the challenges of vessel segmentation in ESD, with the potential to enhance procedural safety and efficacy.

Related Resources & Content

  1. Gastric Cancer — Advancements in Deep Learning Techniques for the Pathological Assessment of Gastric Endoscopic Submucosal Dissection Samples
  2. Evaluation of DNN-Driven Vessel Segmentation for Planning Selective Internal Radiation Therapy
  3. Automatic Registration of Untracked Laparoscopic Ultrasound to Liver CT Through Vessel Segmentation
  4. Endoscopic submucosal dissection for superficial gastrointestinal lesions | ESGE
  5. American Society for Gastrointestinal Endoscopy guideline on endoscopic submucosal dissection for the management of early esophageal and gastric cancers: methodology and review of evidence
  6. Enhanced Transfer Learning for Vascular Segmentation in Laparoscopic Videos Using Adaptive Sensitivity-Fisher Regularization Techniques
  7. Endoscopic submucosal dissection for superficial gastrointestinal lesions | ESGE
  8. Endoscopic submucosal dissection techniques and technology: ESGE Technical Review | ESGE
  9. American Society for Gastrointestinal Endoscopy guideline on endoscopic submucosal dissection for the management of early esophageal and gastric cancers: methodology and review of evidence
  10. Recurrence rates following endoscopic mucosal resection versus endoscopic submucosal dissection for colorectal polyps: a systematic review and meta-analysis of randomized controlled trials | Surgical Endoscopy | Springer Nature Link
  11. Endoscopic submucosal dissection versus endoscopic mucosal resection for colorectal laterally spreading tumors (> 20 mm): a systematic review and meta-analysis | Techniques in Coloproctology | Springer Nature Link
  12. Effect of Prophylactic Colon ESD (Endoscopic Submucosal Dissection) Defect Closure on Post-ESD Outcomes: An International Multi-center Retrospective Study | Digestive Diseases and Sciences | Springer Nature Link
  13. Proper Management and Prevention of Bleeding and Perforation in Endoscopic Submucosal Dissection - ScienceDirect

Original Source(s)

ESD-VesNet: uncertainty-aware vessel segmentation network for endoscopic submucosal dissection with hard negative mining

  • by Mengya Xu, Ming Chen, Zhen Li, Chaoyang Lyu, An Wang, Rulin Zhou, Chuanhao Zhao, Jiaxun Xiang, Tsz Chun Wong, Hossein Farahnaki, Sobhan Zamani Kiasari, Tong Wu, Zimeng Su, Yile Zeng, Ruijing Wen, Xiaohan Shang, Yi Mu, Kezhen Lin, Yidong Zhang, Hongliang Ren

  • July 3, 2026

Related Content

Who Decides When a Doctor Steps Back?

Cognitive decline, credentialing, and why “I’ll know” often isn’t enough.

Impact of Hospice Care Plus Estazolam on Sleep and Psychological Distress in Advanced Gastrointestinal Cancer: A Real-World Observational Study

A novel bimanual endoluminal robotic platform compatible with standard gastroscopes for endoscopic submucosal dissection: preclinical feasibility and ergonomic evaluation