Got Guts: A Chance to Put Liver Biopsy Back to Its Former Glory?

In this episode, coauthors Camilla Venturin and Luca Fabris delve into their editorial titled "Machine Learning Application to Histology for the Study of Cholangiopathies (BiliQML): A Chance to Put Liver Biopsy Back to Its Former Glory?" The episode explores groundbreaking research by Dominick Hellen and colleagues on the limitations of current histological techniques in studying cholangiocytes and the biliary tree.

Historically, the study of these cells has been constrained by outdated and error-prone methodologies, like two-dimensional cell counting or complex three-dimensional imaging that fails to provide reliable quantification. Enter BiliQML, a novel machine learning model that promises to revolutionize this field. This episode breaks down how BiliQML quantifies biliary forms using anti-Keratin 19 antibody-stained whole slide images, providing a far more accurate and scalable approach.

With an impressive F-score of 0.87, the model's application across a variety of cholangiopathy models, including genetic, surgical, toxicological, and therapeutic, showcases its sensitivity and robustness. The episode reveals how this cutting-edge technology opens new doors for both clinical and basic-science researchers in the field of cholangiopathies. Tune in to discover how machine learning is bringing liver biopsy back to the forefront of research.

 

Machine learning application to histology for the study of cholangiopathies (BiliQML): A chance to put liver biopsy back to its former glory?

Camilla Venturin and Luca Fabris

American Journal of Physiology-Gastrointestinal and Liver Physiology 2024 327:6, G733-G736