An open access, machine learning pipeline for high-throughput quantification of cell morphology
Emma M Welter1, Oksana Kosyk1, Anthony S Zannas2
1Department of Psychiatry, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA.
STAR Protocols
|December 17, 2022
Summary
This study introduces a machine-learning protocol for high-throughput cell morphology analysis. The method uses open-access software for accurate measurement of human fibroblasts and other cell types.
Area of Science:
- Cell Biology
- Bioimaging
- Machine Learning Applications
Background:
- Cell morphology serves as a crucial phenotypic marker, influenced by various biological factors.
- Accurate and high-throughput measurement of cell morphology is essential for biological research.
Purpose of the Study:
- To develop and describe a machine-learning-based protocol for high-throughput morphological measurement of human fibroblasts.
- To demonstrate the broad applicability of the protocol to other cell types.
Main Methods:
- Utilized a standard fluorescence microscope.
- Employed open-access software: ilastik for cell body identification, ImageJ/Fiji for image overlay, and CellProfiler for morphological quantification.
- Developed a protocol that overlays cell nuclei with their corresponding cell bodies.
Main Results:
- Successfully established a machine-learning protocol for high-throughput cell morphology analysis.
- The protocol demonstrated effectiveness in measuring human fibroblast morphology.
- The method's reliance on coloration differences and nucleus-cell body overlay suggests broad applicability.
Conclusions:
- The described protocol offers a robust and accessible method for high-throughput cell morphology quantification.
- This approach, leveraging open-access tools, can be adapted for various cell types beyond fibroblasts.
- The study provides a valuable resource for researchers studying cell phenotype and morphology.
