Uncertainty-Driven Vessel Segmentation Network for Endoscopic Submucosal Dissection Incorporating Hard Negative Mining Techniques
-
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
Clinical Scorecard: Uncertainty-Driven Vessel Segmentation Network for Endoscopic Submucosal Dissection Incorporating Hard Negative Mining Techniques
At a Glance
| Category | Detail |
|---|
| Condition | Endoscopic Submucosal Dissection (ESD) |
| Key Mechanisms | AI-assisted real-time detection and segmentation of vessels to minimize intraoperative bleeding risk. |
| Target Population | Patients undergoing ESD for early-stage gastrointestinal cancers. |
| Care Setting | Minimally invasive surgical environments. |
Key Highlights
- Proposes ESD-VesNet for accurate vessel detection and segmentation in ESD procedures.
- Introduces the ESD-Vessel dataset with 2401 annotated vessel frames and 708 hard negative frames.
- Integrates evidential deep learning for uncertainty quantification and hard negative mining.
Guideline-Based Recommendations
Diagnosis
- Utilize AI-assisted tools for improved vessel detection rates during ESD.
Management
- Perform pre-coagulation on potential vessels to prevent intraoperative bleeding.
Monitoring & Follow-up
- Monitor for intraoperative bleeding and manage it promptly to maintain surgical field clarity.
Risks
- Intraoperative bleeding can obscure the surgical field and prolong procedures.
Patient & Prescribing Data
Patients with early-stage gastrointestinal cancers undergoing ESD.
AI tools can enhance vessel detection and reduce complications during ESD.
Clinical Best Practices
- Incorporate AI-assisted vessel segmentation to improve procedural safety.
- Utilize high-quality annotated datasets for training segmentation models.
Related Resources & Content
