Related Experiment Video
Updated: Oct 3, 2025
03:38
Unilateral Lung Volume Analysis Using Micro-CT for Enhanced Assessment of Pulmonary Fibrosis in Preclinical Models
Published on: June 20, 2025
410
FIBER-ML, an Open-Source Supervised Machine Learning Tool for Quantification of Fibrosis in Tissue Sections
Caterina Facchin1, Anais Certain1, Thulaciga Yoganathan1
1Université de Paris, INSERM, Paris Cardiovascular Research Center, Paris, France.
The American Journal of Pathology
|February 20, 2022
Summary
FIBER-ML, an open-source software, accurately quantifies pathologic fibrosis in tissue sections using machine learning. This tool provides a reliable and user-friendly alternative for disease extent assessment in chronic conditions.
Area of Science:
- Biomedical Engineering
- Pathology
- Computational Biology
Background:
- Pathologic fibrosis is a key indicator of chronic disease severity.
- Accurate quantification of fibrosis in tissue sections remains a challenge due to a lack of standardized methods.
- Existing software solutions may lack accessibility or require extensive user input.
Purpose of the Study:
- To evaluate FIBER-ML, a novel, open-source, machine-learning-based freeware for semi-automated fibrosis quantification.
- To compare the performance of FIBER-ML against established quantification tools (ImageJ, inForm) in animal models.
- To assess the reproducibility and user-friendliness of FIBER-ML for fibrosis analysis.
Main Methods:
- FIBER-ML was utilized to quantify fibrosis in sirius red-stained tissue sections from two animal models: stress-induced cardiomyopathy in rats and HIV-induced nephropathy in mice.
- Quantitative results from FIBER-ML were compared with ImageJ (for cardiomyopathy) and inForm (for nephropathy).
- Intra- and inter-operator, as well as inter-software correlations and agreement, were statistically assessed.
Main Results:
- FIBER-ML demonstrated excellent correlations (>0.95) with existing methods in both datasets.
- The software achieved high discriminatory power between pathologic and healthy tissue (<10^-3 and <10^-4).
- Good to moderate agreement was observed for intra-operator (>0.8), inter-operator, and inter-software analyses (>0.7).
Conclusions:
- FIBER-ML provides fast, user-friendly, and reproducible fibrosis quantification in tissue sections.
- This open-source software offers a valuable alternative for researchers needing reliable fibrosis assessment.
- FIBER-ML includes features for quality control and file management, enhancing its utility in research settings.
