Related Experiment Video
Updated: May 20, 2026
17:01
Acquiring Fluorescence Time-lapse Movies of Budding Yeast and Analyzing Single-cell Dynamics using GRAFTS
Published on: July 18, 2013
BudFinder: A Masked Auto-Encoder vision transformer framework for yeast budding detection and lifespan quantification
Phuc Nguyen1, Zahra Mousavi Karimi2, Adrian Layer1
1Department of Molecular Biology, School of Biological Sciences, University of California, San Diego, La Jolla, California, United States of America.
Plos Computational Biology
|May 18, 2026
Summary
This study introduces an automated method for quantifying yeast replicative aging using deep learning. The approach significantly reduces the need for manual data annotation, accelerating aging research.
Area of Science:
- Gerontology
- Cell Biology
- Biophysics
Background:
- Replicative aging in yeast is crucial for understanding aging.
- Automating lifespan quantification from microscopy is needed.
- Current methods are labor-intensive and require manual cell division counting.
Purpose of the Study:
- To develop a versatile image analysis framework for accurate yeast cell division event detection.
- To reduce the dependency on large, manually annotated datasets for deep learning models.
- To improve the efficiency and accuracy of replicative lifespan quantification.
Main Methods:
- Utilized a Masked Autoencoder for self-supervised pretraining on unlabeled yeast cell images (~250K).
- Trained a transformer model for division event detection using limited annotated data.
- Focused on direct budding event identification, avoiding indirect heuristics like cell area changes.
Main Results:
- Achieved accurate detection of yeast cell division events during replicative aging.
- Significantly reduced the requirement for annotated data (fewer than 50 mother cells) compared to previous methods.
- Demonstrated high detection accuracy through self-supervised learning and direct budding event identification.
Conclusions:
- The developed framework enables efficient and accurate automated quantification of yeast replicative lifespan.
- Self-supervised learning drastically lowers annotation burden, making deep learning more accessible for aging research.
- This method advances high-throughput analysis of cellular aging processes.
