Caenorhabditis elegans segmentation using texture-based models for motility phenotyping.

Area of Science:

• Biophysics
• Robotics
• Developmental Biology

Background:

• Motility phenotyping of Caenorhabditis elegans is crucial for genetic and robotics research.
• Automatic segmentation of C. elegans in diverse environments remains a significant challenge.
• Existing methods struggle with complex and dynamic visual cues in image sequences.

Purpose of the Study:

• To develop a novel automatic segmentation method for C. elegans.
• To address the bottleneck in extracting motility phenotypes from image sequences.
• To improve segmentation accuracy in challenging and dynamic locomotive environments.

Main Methods:

• Introduction of a texture factor model (TFM) integrating intensity- and texture-based features.
• Utilizing a probabilistic framework with Markov random field for segmentation refinement.
• Employing approximate inference techniques and informative priors for coherent sequence segmentation.

Main Results:

• TFM demonstrates comparable performance to existing methods in static environments.
• TFM achieves state-of-the-art segmentation in dynamic environments, including wet granular media.
• Successfully computed nematode skeletons from segmentations for subsequent motility phenotype extraction.

Conclusions:

• The TFM provides an effective solution for C. elegans segmentation in challenging environments.
• This method facilitates advanced motility phenotype analysis.
• Enables wider application of C. elegans as a model organism in diverse research areas.