Related Experiment Video
Updated: Jul 8, 2026
03:56
A Two-Step Method for Percutaneous Transhepatic Choledochoscopic Lithotomy
Published on: September 13, 2022
2.4K
Multiple Field-of-View Based Attention Driven Network for Weakly Supervised Common Bile Duct Stone Detection.
Ya-Han Chang1, Meng-Ying Lin2, Ming-Tsung Hsieh2
1Department of Computer Science and EngineeringNational Chung Hsing University Taichung 402202 Taiwan.
Summary
Detecting common bile duct (CBD) stones using CT scans is challenging. A new deep learning model, MFADNet, accurately locates these stones with image-level labels, aiding physician diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Gastroenterology
Background:
- Common bile duct (CBD) stones pose life-threatening risks.
- Detecting small, distal CBD stones in CT scans is difficult.
Purpose of the Study:
- To develop a weakly-supervised deep learning method for detecting CBD stones in CT scans.
- To reduce the burden of detailed labeling for medical professionals.
Main Methods:
- Proposed a multiple field-of-view based attention driven network (MFADNet).
- Employed a multiple field-of-view encoder and an attention-driven decoder with a classification network.
- Utilized four losses (foreground, background, consistency, classification) for end-to-end training.
Main Results:
- MFADNet accurately classifies and locates CBD stones.
- Demonstrated superior performance compared to state-of-the-art weakly-supervised methods.
Conclusions:
- MFADNet offers a novel, weakly-supervised approach for CBD stone detection from CT scans.
- The method assists physicians in the automatic diagnosis of CBD stone-related diseases.
