Related Experiment Video
Updated: Aug 14, 2026
07:44
Endoscopic Ultrasound-Guided Biliary Drainage: Endoscopic Ultrasound-Guided Hepaticogastrostomy in Malignant Biliary Obstruction
Published on: March 25, 2022
Deep Learning-Based Computer-Aided Detection and Diagnosis System for Malignant Biliary Stricture (With Video)
Qingyu Tang1, Sanping Zhou2, Zizhan Tang3
1Department of Hepatobiliary Surgery, The First Affiliated Hospital of Xi'an Jiaotong University, Xi'an 710061, China.
Cancers
|August 13, 2026
Summary
A new deep learning system aids in diagnosing malignant biliary strictures using digital cholangioscopy. The system shows promise, but further real-world testing is needed for broader application.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Gastroenterology
Background:
- Malignant biliary stricture (MBS) diagnosis is challenging, even with digital single-operator cholangioscopy (DSOC).
- Accurate MBS detection is crucial for timely treatment and improved patient outcomes.
Purpose of the Study:
- To develop and evaluate a deep learning (DL)-based computer-aided detection (CADe) and diagnosis (CADx) system for MBS assessment using DSOC.
- To improve the accuracy and efficiency of MBS diagnosis.
Main Methods:
- A retrospective multicenter study utilized a DL system with YOLOv11 for feature localization and ResNet-18 for classification.
- The system was trained and internally validated on 149 patients and externally evaluated on 25 patients from independent centers.
Main Results:
- The CADe component achieved high precision (92.0%) and recall (87.0%) in localizing key features.
- The CADx system demonstrated strong performance with an AUC of 0.960 (internal) and 0.881 (patient-level external).
- External validation showed a patient-level sensitivity of 85.7% and specificity of 94.4%.
Conclusions:
- The developed two-stage DL system effectively combines feature localization and interpretable classification for MBS assessment.
- While promising, the study highlights the need for prospective multicenter trials and live-procedure evaluation to confirm generalizability due to limitations in the external cohort size and sensitivity.
