Machine learning driven semi-automated framework for yeast sporulation efficiency quantification using ilastik...

Area of Science:

• * Microbiology and Genetics
• * Bioimage Analysis

Background:

• * Accurate quantification of yeast sporulation efficiency is crucial for genetic studies.
• * Manual counting is labor-intensive and prone to subjective errors.
• * Existing deep learning tools may not be universally applicable or adaptable.

Purpose of the Study:

• * To develop an automated, robust, and accessible pipeline for yeast sporulation efficiency quantification.
• * To reduce processing time and subjective bias associated with manual counting.
• * To provide a reliable alternative for diverse genetic backgrounds and spore morphologies.

Main Methods:

• * Utilized ilastik for texture-feature optimization to segment sporulating yeast cells.
• * Employed Fiji for optimized image processing and spore quantification within segmented cells.
• * Implemented automated classification of dyads, triads, and tetrads with manual quality control checkpoints.

Main Results:

• * Achieved 93.4% agreement with manual counting (ICC = 0.94).
• * Reduced processing time by 68% (P < 0.001).
• * Demonstrated consistent performance across Hsp82 phosphorylation mutants and diverse genetic backgrounds.

Conclusions:

• * The developed pipeline offers a reproducible and precise alternative to manual yeast sporulation quantification.
• * The modular design allows for adjustable parameters and compatibility with various imaging datasets and markers.
• * This method balances throughput and accuracy, making it suitable for standard laboratory microscopy.