CellSNAP: a fast, accurate algorithm for 3D cell segmentation in quantitative phase imaging

Area of Science:

• Biophysics
• Cell Biology
• Image Analysis

Background:

• Quantitative Phase Imaging (QPI) provides label-free cell morphology and dynamics analysis.
• Current QPI analysis pipelines lack well-developed 3D cell segmentation tools.
• Existing methods face limitations with photobleaching, phototoxicity, and contrast agent variability.

Purpose of the Study:

• To develop a novel algorithm for 3D cell segmentation in QPI.
• To address the critical step of analyzing raw 3D tomograms in QPI.
• To enable high-throughput analysis of QPI data.

Main Methods:

• The Cell Segmentation via Novel Algorithm for Phase Imaging (CellSNAP) algorithm was developed.
• It employs a gemstone extraction analogy: coarse 3D extrusion followed by refined segmentation using cell continuity.
• The algorithm processes 2D segmented masks and leverages continuity across 3D stacks.

Main Results:

• CellSNAP achieves segmentation in under 2 seconds per cell on a single-core processor.
• It demonstrates robustness in handling clumped cells and interferogram drifts, outperforming AI-based tools.
• The algorithm shows minimal differences in dry mass (5%) and volume (8%) compared to gold standards.

Conclusions:

• CellSNAP is faster, less memory-intensive, and more robust than existing 3D QPI segmentation methods.
• Its rule-based approach eliminates the need for extensive training data, facilitating broader QPI adoption.
• This tool is envisioned to accelerate high-throughput QPI analysis, overcoming a key bottleneck.