A bioimage informatics approach to automatically extract complex fungal networks

Area of Science:

• Mycology
• Computational Biology
• Image Analysis

Background:

• Fungi form complex, adaptive mycelial networks for resource acquisition.
• Analyzing these networks is crucial for understanding fungal ecology and development.
• Manual analysis of high-throughput fungal network images is infeasible due to scale and complexity.

Purpose of the Study:

• To develop and evaluate a high-throughput automated image analysis approach for characterizing fungal networks.
• To enable robust detection and graph-based representation of complex curvilinear fungal structures.

Main Methods:

• Utilized Phase Congruency Tensors and watershed segmentation for image processing.
• Applied the method to complex images of saprotrophic fungal networks with millions of edges.
• Developed a graph-based representation for network analysis.

Main Results:

• The automated approach provides fast and robust detection of fungal network branches.
• Successfully characterized complex fungal networks with high throughput.
• Enabled graph-based representation for detailed network analysis.

Conclusions:

• The developed image analysis method is effective for high-throughput characterization of fungal networks.
• This tool facilitates deeper understanding of fungal network architecture and adaptation.
• The approach overcomes challenges posed by variable contrast and heterogeneous substrates.