Label-free rapid antimicrobial susceptibility testing with machine-learning based dynamic holographic laser speckle...

Area of Science:

• Biomedical Optics
• Microbiology
• Machine Learning

Background:

• Antimicrobial resistance (AMR) necessitates rapid antimicrobial susceptibility testing (AST).
• Traditional AST methods are time-consuming, delaying critical treatment.
• Urgent need for faster diagnostic tools to guide antimicrobial stewardship.

Purpose of the Study:

• To develop and validate a rapid, label-free AST method using dynamic holographic laser speckle imaging (DhLSI).
• To assess bacterial susceptibility to antibiotics within hours, not days.
• To leverage machine learning for enhanced AST performance.

Main Methods:

• Implementation of a dynamic holographic laser speckle imaging (DhLSI) system with a reference beam.
• Utilizing artificial neural networks (ANN) to analyze dynamic speckle patterns.
• Testing bacterial susceptibility at concentrations as low as 10^3 CFU/mL.

Main Results:

• DhLSI system achieved results consistent with standard AST concentrations (10^5 CFU/mL).
• Accurate bacterial susceptibility determination within 2-3 hours.
• Consistent performance across different bacterial strains and incubation days.

Conclusions:

• The DhLSI system offers a rapid and sensitive platform for AST.
• This technology can significantly improve early-stage infection diagnosis and antimicrobial stewardship.
• Potential for clinical implementation to combat the global challenge of AMR.