Related Experiment Video
Updated: Jun 28, 2025
07:35
Determining Bile Duct Density in the Mouse Liver
Published on: April 30, 2019
6.9K
BiliQML: a supervised machine-learning model to quantify biliary forms from digitized whole slide liver
Dominick J Hellen1, Meredith E Fay2,3, David H Lee1
1Division of Pediatric Gastroenterology, Hepatology, and Nutrition, Department of Pediatrics, Children's Healthcare of Atlanta and Emory University School of Medicine, Atlanta, Georgia, United States.
Summary
Researchers developed BiliQML, a machine-learning tool to quantify biliary forms in liver images. This novel platform overcomes limitations of current methods, offering a robust and scalable solution for studying cholangiopathies.
Area of Science:
- Hepatology and Biliary Research
- Computational Pathology
- Biomedical Imaging Analysis
Background:
- Quantitative analysis of cholangiocytes and the biliary tree is crucial for understanding liver development and disease.
- Existing histological and imaging methods for biliary tree analysis are limited by accuracy, architectural context, and technical challenges.
Purpose of the Study:
- To introduce BiliQML, a supervised machine-learning model for quantifying biliary forms in anti-keratin 19 antibody-stained whole slide liver images.
- To address the quantitative gaps in current methodologies for biliary tree research.
Main Methods:
- Development of a supervised machine-learning model (BiliQML) trained on 5,019 researcher-labeled biliary forms.
- Feature selection and algorithm optimization achieved an F-score of 0.87.
- Validation of BiliQML across seven diverse cholangiopathy models (genetic, surgical, toxicological, and therapeutic).
Main Results:
- BiliQML accurately quantified distinct biliary forms across various cholangiopathy models.
- The platform revealed significant biological and pathological differences between models.
- Demonstrated high sensitivity, robustness, and scalability for biliary form quantification.
Conclusions:
- BiliQML is the first comprehensive machine-learning platform for analyzing biliary forms in histopathological images.
- It adds essential morphologic context to immunofluorescence-based histology.
- Provides a novel, powerful tool for researchers studying cholangiopathies and biliary tract disorders.
