Related Experiment Video
Updated: Jan 9, 2026
08:59
Morphology-Based Distinction Between Healthy and Pathological Cells Utilizing Fourier Transforms and Self-Organizing Maps
Published on: October 28, 2018
7.5K
A morphology-based machine learning model for scoring epithelial-mesenchymal plasticity using organelle dynamics
Justin Slager1, Francesca Gatto1, Benjamin Frey2
1Division of Pathology, Department of Laboratory Medicine, Karolinska Institutet, Stockholm, Sweden.
Communications Biology
|December 10, 2025
Summary
We developed a machine learning method to quantify epithelial-mesenchymal transition (EMT) states by analyzing organelle dynamics. This approach accurately scores EMT in various cancer cells, offering a new tool for drug discovery and therapy resistance research.
Area of Science:
- Cell Biology
- Cancer Research
- Computational Biology
Background:
- Epithelial-mesenchymal transition (EMT) is crucial for cancer progression and therapy resistance.
- Quantifying EMT states, including hybrid phenotypes, is challenging but vital for therapeutic strategies.
- Existing methods lack the ability to capture the full spectrum of EMT dynamics.
Purpose of the Study:
- To develop a novel morphology-based machine learning approach to quantify EMT states.
- To establish a scalable framework for assessing epithelial-mesenchymal plasticity.
- To enable drug discovery and identify resistance-overcoming strategies.
Main Methods:
- Utilized the Cell Painting assay and high-throughput microscopy to profile organelle dynamics.
- Trained a histogram gradient boosting classifier on time-course data of TGF-β1-induced EMT.
- Validated the model's performance across different datasets and cell types.
Main Results:
- The machine learning model accurately quantified EMT kinetics, hybrid states, and mesenchymal-epithelial transition (MET).
- Demonstrated cross-species, cross-inducer, and cross-cancer applicability in human breast and lung cancer cells.
- Established organelle morphology profiling as a robust method for assessing epithelial-mesenchymal plasticity.
Conclusions:
- Organelle morphology profiling provides a scalable and accurate method for quantifying EMT.
- This approach can be applied to various cancer types and inducers, including hypoxia.
- The developed framework offers a promising platform for therapeutic interventions against EMT-associated cancer progression and resistance.
