Towards automated cellular image segmentation for RNAi genome-wide screening

Area of Science:

• Cell Biology
• Bioinformatics
• Computational Biology

Background:

• Rho GTPases are crucial for cell morphology, development, migration, and cancer.
• Genome-wide RNA interference (RNAi) screening is a powerful tool for discovering genes involved in cellular processes.
• Manual analysis of high-content imaging data from RNAi screens is a significant bottleneck.

Purpose of the Study:

• To develop an automated image analysis method for high-throughput cell phenotype classification in genome-wide RNAi screens.
• To identify novel genes and pathways regulating Rho protein activity and cell morphology.

Main Methods:

• A two-step cell segmentation approach using DNA-channel information, dissimilarity functions, and polygon estimation.
• Fuzzy c-means based multi-threshold segmentation, sharpening, and marker-controlled watershed for cell isolation.
• Voronoi diagrams for correcting segmentation errors, followed by feature extraction and K-nearest neighbor (KNN) classification.

Main Results:

• The proposed automated image analysis pipeline successfully segments and classifies cell phenotypes from complex imaging data.
• The method effectively handles challenges like over-segmentation and touching/overlapping cells.

Conclusions:

• The developed image analysis approach significantly accelerates the discovery of genes affecting cell phenotypes in large-scale RNAi screens.
• This method facilitates the identification of novel Rho effector genes and enhances our understanding of cell morphology regulation.