Related Experiment Video
Updated: May 16, 2025
10:10
Measurement of Liver Stiffness Using Atomic Force Microscopy Coupled with Polarization Microscopy
Published on: July 20, 2022
3.7K
Characterization of fibrotic liver tissue microstructure for predicting shear wave speed variability: a
Emily J Miller1, Yongmei M Jin1, Jingfeng Jiang2
1Department of Materials Science and Engineering, Michigan Technological University, Houghton, MI, United States of America.
Physics in Medicine and Biology
|April 4, 2025
Summary
Machine learning links liver fibrosis microstructure to shear wave speed (SWS) variability. While percent inclusion correlates with SWS, it alone cannot predict fibrosis, necessitating analysis of both mean SWS and SWS-STD for accurate tissue characterization.
Area of Science:
- Biomedical Engineering
- Medical Physics
- Computational Biology
Background:
- Liver fibrosis assessment is crucial for disease management.
- Shear wave speed (SWS) variability offers insights into tissue microstructure.
- Machine learning (ML) can potentially link microstructural features to SWS.
Purpose of the Study:
- To establish a link between simulated fibrotic liver tissue microstructure and SWS variability using ML.
- To identify key microstructural features influencing SWS variability.
- To determine if SWS variability can infer underlying tissue microstructure.
Main Methods:
- Simulated fibrotic liver tissues using biphasic random fields.
- Characterized microstructure via spatial pattern distribution analysis.
- Implemented ML to identify spatial characteristic (SC) features and predict SWS variability.
Main Results:
- Percent inclusion correlated with SWS estimates but was not a sole predictive factor.
- No single SC feature accurately predicted SWS estimates in simulated tissues.
- Top ML-identified features, though varying in name, were highly correlated across iterations.
Conclusions:
- Percent inclusion alone is insufficient for predicting mean SWS or SWS-STD in fibrotic liver tissue.
- Both mean SWS and SWS-STD provide unique, valuable information about tissue microstructure.
- Simulated fibrotic liver pathology (sFLP) models show consistency with real-world data, validating their representativeness.
