Morphology-guided graph search for untangling objects: C. elegans analysis.

Area of Science:

• Computational Biology
• Bioimage Analysis
• High-Throughput Screening

Background:

• Untangling clustered Caenorhabditis elegans (C. elegans) is crucial for accurate analysis in high-throughput screening.
• Existing methods struggle with dense worm clusters, impacting data reliability.

Purpose of the Study:

• To develop a novel computational approach for automated extraction of individual C. elegans from cluttered clusters.
• To improve accuracy and efficiency in C. elegans phenotyping.

Main Methods:

• Representing worm clusters as sparse directed graphs based on morphological properties.
• Employing a graph search algorithm to identify individual worm paths while minimizing overlap.
• Utilizing a low-dimensional feature space for worm pose estimation derived from isolated worm training data.

Main Results:

• Successfully untangled C. elegans clusters in 236 microscopy images.
• Achieved high worm detection accuracy, with each image containing approximately 15 worms.
• Demonstrated the effectiveness of the morphological and graph-based approach.

Conclusions:

• The proposed method offers a robust solution for C. elegans cluster untangling in bioimage analysis.
• This technique enhances the reliability of automated phenotyping in high-throughput screening experiments.
• The graph-based approach effectively leverages morphological features for accurate object extraction.