Advancing Hydrogel-Based 3D Cell Culture Systems: Histological Image Analysis and AI-Driven Filament Characterization

Area of Science:

• Biomaterials Science
• Materials Engineering
• Artificial Intelligence

Background:

• Machine learning analyzes images by training algorithms to recognize patterns, with broad applications including medicine and automation.
• Histological cross-sections reveal tissue layers crucial for microscopic analysis.
• Artificial intelligence (AI) offers potential for interpreting complex biological and material structures.

Purpose of the Study:

• To utilize Google's Teachable Machine platform for applying AI in interpreting histological cross-section images of hydrogel filaments.
• To assess the efficacy of AI in differentiating between various states of hydrogel filament structures.

Main Methods:

• 3D hydrogel filaments were fabricated using varying ratios of sodium alginate and gelatin polymers with a cross-linking agent.
• Filaments were subjected to stretching until rupture using an extensometer.
• Cross-sections of stretched and unstretched filaments were prepared, stained (hematoxylin and eosin), and imaged for AI training and prediction via Teachable Machine.

Main Results:

• Over 600 histological cross-section images were collected and cataloged.
• AI models successfully differentiated between stretched and unstretched hydrogel filament images.
• While effective for state differentiation, AI exhibited some limitations in distinguishing subtle variations among different hydrogel formulations.

Conclusions:

• The AI-powered image prediction tool, implemented via Teachable Machine, proved efficient for classifying hydrogel filaments as stretched or unstretched.
• This approach demonstrates a viable strategy for the AI-driven analysis of biopolymeric hydrogel histology.