Artificial Intelligence-Driven Image Analysis of Bacterial Cells and Biofilms

Area of Science:

• Microbiology
• Materials Science
• Computer Science

Background:

• Bacterial biofilms, particularly sulfate-reducing bacteria like Desulfovibrio alaskensis G20, contribute to microbiologically influenced corrosion (MIC) on metal surfaces.
• Understanding the structural and geometric properties of biofilms is crucial for developing effective corrosion prevention strategies.
• Manual analysis of biofilm cell geometry from scanning electron microscopy (SEM) images is time-consuming and labor-intensive.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) framework for automated measurement of bacterial cell structural features from SEM images.
• To quantify the geometric properties of Desulfovibrio alaskensis G20 biofilms to understand their adaptation to mild steel surfaces.
• To accelerate the analysis of biofilm phenotypes for improved insights into MIC.

Main Methods:

• Adaptation of two deep learning models: a deep convolutional neural network (DCNN) for semantic segmentation and a Mask Region-based Convolutional Neural Network (Mask R-CNN) for instance segmentation of bacterial cells.
• Integration of segmentation models with the moment invariants approach to extract geometric characteristics of individual bacterial cells.
• Utilizing Desulfovibrio alaskensis G20 grown on mild steel as a model system for sulfate-reducing bacteria.

Main Results:

• The AI framework successfully segmented bacterial cells, enabling the measurement of their geometric properties.
• Numerical studies demonstrated significant speed improvements: Mask R-CNN was 227x faster, and DCNN was 70x faster than manual measurement.
• The automated approach provides a rapid and efficient method for characterizing biofilm phenotypes.

Conclusions:

• The developed AI framework offers a highly efficient and automated solution for analyzing bacterial biofilm structures from microscopy images.
• This advancement facilitates a deeper understanding of biofilm adaptation and its role in microbiologically influenced corrosion.
• The accelerated analysis holds potential for improved design of corrosion prevention methods in various industrial applications.