On the Variability in Cell and Nucleus Shapes.

Area of Science:

• Cellular Biology
• Biophysics
• Computational Imaging

Background:

• Cell morphology significantly influences cell function, with abnormalities often detectable in cell and nuclear shape.
• Characterizing 3D cell and nuclear shapes from confocal microscopy data presents challenges in reconstruction and comparative analysis.

Purpose of the Study:

• To develop and validate algorithms for segmenting 3D cell and nucleus surfaces from confocal images.
• To create a reversible vector-based representation for quantifying and comparing cell and nuclear shapes.

Main Methods:

• Utilized 3D active contours and a 3D condensed-attention UNet for segmenting cells and nuclei.
• Employed a reversible spherical transform to convert segmented surfaces into a vector-based shape representation.
• Applied principal component analysis (PCA) to the vector representation to identify modes of shape variability.

Main Results:

• Successfully segmented 3D cell and nucleus surfaces, enabling a novel vector-based shape characterization.
• PCA revealed that scaling and flattening are primary modes of variability in cell and nucleus shapes.
• The vector representation was benchmarked against a mechanical model of nucleus morphology.

Conclusions:

• The developed vector-space representation provides a physically interpretable method for analyzing cell and nuclear shape variability.
• This approach facilitates the understanding of mechanical models and the identification of molecular mechanisms underlying shape changes.