Calibration for Quantitative Chemical Analysis in IR Microscopic Imaging

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Biotechnology

Background:

• Infrared spectroscopy offers fast, low-cost analysis of macroscopic samples.
• Quantitative analysis of microscopic infrared spectral data has been limited.
• Spatially resolved chemical information from infrared microspectroscopy is challenging.

Purpose of the Study:

• To develop a deep learning-based method for quantitative chemical analysis of hyperspectral infrared images.
• To enable calibration transfer from macroscopic infrared spectroscopy models to microscopic pixel spectra.
• To overcome limitations in spatially resolved quantitative analysis using infrared microspectroscopy.

Main Methods:

• A deep learning-based calibration transfer technique was developed.
• Infrared microspectroscopic data was adapted to the domain of macroscopic spectra.
• Regression models established for bulk measurements were applied to hyperspectral images.

Main Results:

• The method successfully adapted macroscopic calibration models for use with microscopic spectra.
• Quantitative chemical analysis was achieved directly from infrared microspectroscopic measurements.
• The approach was validated using oleaginous filamentous fungi, correlating with gas chromatography lipid profiles and glucosamine content.

Conclusions:

• Deep learning enables calibration transfer for quantitative infrared microspectroscopy.
• This method opens new possibilities for spatially resolved chemical analysis in imaging.
• The approach is valuable for analyzing biological samples like fungi.