Microscope-Based Automated Quantification of Liver Fibrosis in Mice Using a Deep Learning Algorithm

In preclinical studies that involve animal models for hepatic fibrosis, accurate quantification of the fibrosis is of utmost importance. The use of digital image analysis based on deep learning artificial intelligence (AI) algorithms can facilitate accurate evaluation of liver fibrosis in these models. In the present study, we compared the quantitative evaluation of collagen proportionate area in the carbon tetrachloride model of liver fibrosis in the mouse by a newly developed AI algorithm to the semiquantitative assessment of liver fibrosis performed by a board-certified toxicologic pathologist. We found an excellent correlation between the 2 methods of assessment, most evident in the higher magnification (x40) as compared to the lower magnification (x10). These findings strengthen the confidence of using digital tools in the toxicologic pathology field as an adjunct to an expert toxicologic pathologist.

Automated summary

The study compared the quantitative assessment of liver fibrosis using AI algorithms with semiquantitative assessment by a pathologist, showing excellent correlation at higher magnification levels.