Automated classification of bacterial particles in flow by multiangle scatter measurement and support vector machine...

Area of Science:

• Biophotonics
• Analytical Chemistry
• Microbiology

Background:

• Biological microparticles scatter light, creating complex patterns influenced by physical properties.
• Standard flow cytometry uses limited scatter angles, hindering separation of similar cells.
• Existing methods for full scatter pattern analysis require custom-built flow cytometers.

Purpose of the Study:

• To develop a pattern-recognition approach for classifying particles using discrete scatter patterns.
• To determine if limited scatter angles and axial light loss are sufficient for cell population separation.
• To enable the use of enhanced scatter detection with existing flow cytometry instruments.

Main Methods:

• Applied pattern-recognition techniques, specifically a support vector machine (SVM), to classify particles.
• Collected discrete scatter patterns at five different angles, along with axial light loss measurements.
• Utilized an analytical model of laser beam scatter by bacterial cells to optimize sensor placement.

Main Results:

• Demonstrated that five scatter angles and axial light loss provide sufficient information for particle classification.
• Achieved high success rates (68-99%) in recognizing various bacteria based on their scatter patterns.
• Showcased the potential for retrofitting existing flow cytometers with an enhanced scatter detector.

Conclusions:

• Pattern recognition applied to discrete scatter patterns offers a viable method for bacterial identification.
• This approach enhances flow cytometry capabilities without necessitating custom-built instrumentation.
• The method provides a cost-effective and efficient way to differentiate cell populations.