Optimized digital workflow for pathologist-grade evaluation in bleomycin-induced pulmonary fibrosis mouse model.
Toshiki Goto1, Akira Sano2, Shinichi Onishi3
1Research Division, Chugai Pharmaceutical Co., Ltd., 216 Totsuka-cho, Totsuka-ku, Yokohama-shi, Kanagawa, 244-8602, Japan.
Scientific Reports
|January 20, 2025
Summary
Deep learning models now accurately assess lung fibrosis in the bleomycin-induced pulmonary fibrosis mouse model (BLM model). This AI approach improves efficiency and reproducibility for developing new idiopathic pulmonary fibrosis (IPF) treatments.
Area of Science:
- Pulmonary Medicine
- Computational Pathology
- Drug Discovery
Background:
- Idiopathic pulmonary fibrosis (IPF) is a fatal lung disease with unknown causes.
- The bleomycin-induced pulmonary fibrosis mouse model (BLM model) is crucial for evaluating IPF therapies.
- Current histopathological analysis of lung fibrosis in the BLM model is time-consuming and subjective.
Purpose of the Study:
- To develop an efficient and reproducible workflow for evaluating lung fibrosis in the BLM model.
- To create deep learning models for accurate fibrosis grading.
- To enhance the identification of potential drug candidates for IPF.
Main Methods:
- Generation of deep learning models for grading lung fibrosis.
- Utilizing complex image patterns and qualitative factors (e.g., collagen texture and distribution).
- Comparison of model accuracy against expert pathologist evaluations.
Main Results:
- Deep learning models achieved accuracy comparable to human pathologists.
- The developed models reduce inter- and intra-observer variations.
- The approach offers higher granularity and reproducibility in fibrosis assessment.
Conclusions:
- Deep learning-based fibrosis grading streamlines drug development for pulmonary fibrosis.
- This method enhances the evaluation of the BLM model for IPF research.
- The AI approach has the potential to identify novel therapeutic targets for IPF.
