In situ biological particle analyzer based on digital inline holography.

Area of Science:

• Biotechnology
• Microscopy
• Machine Learning

Background:

• Accurate in situ measurements of biological microparticles are vital for scientific research and industrial applications.
• Current diagnostic methods lack the timeliness, accuracy, and detailed information required for effective analysis.
• Challenges exist in real-time monitoring of microparticles like plankton and yeast cells.

Purpose of the Study:

• To develop a novel, real-time, in situ method for analyzing biological microparticles.
• To enhance diagnostic capabilities for applications such as harmful algal bloom detection and fermentation monitoring.
• To leverage machine learning with digital inline holography for improved particle analysis.

Main Methods:

• Utilized digital inline holography (DIH) combined with machine learning (ML).
• Developed a customized YOLOv5 architecture for detecting and classifying small biological particles.
• Applied the ML-assisted DIH method to analyze plankton species and yeast cells with varying metabolic states and strains.

Main Results:

• Achieved high accuracy in detecting and classifying 10 plankton species, with significantly reduced processing time.
• Successfully differentiated yeast cells across four metabolic states and two strains.
• Demonstrated accurate detection and differentiation of cellular and subcellular features linked to metabolic states and strains.

Conclusions:

• The ML-driven DIH approach provides a sensitive and versatile tool for real-time, in situ analysis.
• This method offers a significant improvement over existing techniques for biological microparticle diagnostics.
• The technology is suitable for scalable, distributive deployment in scientific research and industrial manufacturing.