Detailed Protocol for Segmentation and Quantification of Overlapping Prospore Membranes using DeMemSeg

Area of Science:

• Cell Biology
• Biophysics
• Microscopy

Background:

• Quantitative analysis of biological membrane morphology is crucial for understanding cellular processes.
• Manual annotation is laborious and subjective, while automated methods struggle with overlapping structures in 2D microscopy images.

Purpose of the Study:

• To develop a robust and automated method for high-fidelity instance segmentation and quantitative analysis of overlapping prospore membranes (PSMs).
• To enable accurate and reproducible segmentation of individual, overlapping membrane structures from 2D maximum intensity projections (MIPs).

Main Methods:

• A step-by-step protocol involving synchronous sporulation induction, 3D fluorescence image acquisition, and conversion to 2D MIPs.
• Generation of a custom-annotated dataset using a semi-automated pipeline and a CellPose model for single-cell isolation.
• Training and application of a mask R-CNN-based model, DeMemSeg, for instance segmentation.

Main Results:

• DeMemSeg achieves high-fidelity instance segmentation of individual, overlapping PSMs from 2D MIPs.
• The protocol enables extraction of morphological parameters (e.g., length, roundness) for quantitative phenotyping.
• Accurate differentiation between wild-type and mutant yeast strains based on membrane morphology.

Conclusions:

• The DeMemSeg framework provides an objective, efficient, and scalable solution for analyzing complex membrane morphologies.
• This method facilitates detailed quantitative analysis of dynamic membrane structures, particularly yeast PSMs.
• Offers a significant advancement over manual annotation and conventional automated methods for biological membrane analysis.