Microbial Identification Through Multispectral Infrared Imaging of Colonies: A New Type of Morpho-Spectral...

Area of Science:

• Microbiology
• Spectroscopy
• Imaging Technology

Background:

• Accurate and rapid microbial identification is crucial for diagnostics and research.
• Current methods like Fourier-transform infrared (FTIR) spectroscopy can be time-consuming or require sample preparation.

Purpose of the Study:

• To introduce and validate a novel microbial identification technique using Direct Frequency Infrared (DFIR) multispectral imaging.
• To assess the performance of DFIR imaging combined with machine learning for classifying microbial colonies.

Main Methods:

• Utilized Quantum Cascade Laser (QCL) light sources and a microbolometer array for lensless multispectral imaging.
• Captured optical fingerprints combining morphological and spectral data from microbial colonies across 9 mid-infrared wavelengths.
• Developed a proof-of-concept database with 10 strains from 8 species and 2253 colonies.

Main Results:

• Machine learning classification achieved up to 94.4% ± 1.6% accuracy in identifying microbial colonies.
• High classification performance was maintained even when reducing the number of wavelengths to 4.
• The DFIR system successfully discriminated between closely related microbial strains.

Conclusions:

• Direct Frequency Infrared (DFIR) multispectral imaging is a viable, label-free method for microbial identification.
• The developed system demonstrates robustness and high accuracy, outperforming existing FTIR imaging systems in speed and simplicity.
• This technology holds promise for applications requiring rapid and efficient microbial analysis.