Related Experiment Video
Updated: Dec 11, 2025
12:48
A Time-lapse, Label-free, Quantitative Phase Imaging Study of Dormant and Active Human Cancer Cells
Published on: February 16, 2018
7.7K
Automated interpretation of time-lapse quantitative phase image by machine learning to study cellular dynamics during
Lenka Strbkova1,2, Brittany B Carson3, Theresa Vincent3,4
1Brno Univ. of Technology, Czech Republic.
Journal of Biomedical Optics
|August 20, 2020
Summary
Incorporating time-lapse imaging with digital holographic microscopy (DHM) significantly improves machine learning classification of dynamic cellular processes like epithelial-mesenchymal transition (EMT), boosting accuracy by nearly 9%. This approach enhances live cell monitoring and automated analysis.
Area of Science:
- Biophysics
- Cell Biology
- Microscopy
Background:
- Machine learning is increasingly used for microscopic data classification.
- Time-resolved live-cell imaging is crucial for detecting complex cellular dynamics.
- Temporal information can enhance classification specificity and accuracy.
Purpose of the Study:
- To develop a cell classification methodology using time-lapse quantitative phase images (QPIs) from digital holographic microscopy (DHM).
- To improve the classification performance of dynamic cellular processes.
Main Methods:
- Utilized time-lapse QPIs from DHM over a 48-hour period to study epithelial-mesenchymal transition (EMT).
- Extracted novel features representing dynamic cell behavior.
- Classified distinct EMT phenotypes using supervised machine learning algorithms and compared results with single-time-point classifications.
Main Results:
- Incorporating temporal information improved cell phenotype classification accuracy during EMT by nearly 9% compared to single-time-point analysis.
- Demonstrated the potential of DHM for monitoring cellular morphological changes over time.
Conclusions:
- The proposed DHM-based, time-lapse approach enhances automated monitoring of live cell behavior.
- This methodology can be developed into a tool for high-throughput, automated analysis of unique cell behaviors.
