Multi-modal image sharpening in fourier transform infrared (FTIR) microscopy

Area of Science:

• Spectroscopy
• Microscopy
• Biomedical Imaging

Background:

• Mid-infrared Spectroscopic Imaging (MIRSI) offers molecular specificity but is limited by diffraction-limited resolution.
• Longer wavelengths (2.5-12.5 μm) essential for molecular fingerprinting (6-12 μm) restrict resolution to 3-6 μm.
• This resolution limit is a significant barrier in biomedical and materials science applications.

Purpose of the Study:

• To overcome the spatial resolution limitations inherent in MIRSI.
• To develop a method for augmenting infrared images with higher-resolution visible microscopy data.
• To generate fused hyperspectral datasets with both high spatial resolution and molecular contrast.

Main Methods:

• An unsupervised curvelet-based image fusion algorithm was developed.
• The method fuses mid-infrared spectroscopic images with label-free visible light microscopy images.
• The technique was demonstrated on breast and ovarian tumor biopsy samples.

Main Results:

• The proposed fusion method successfully generated hyperspectral datasets with enhanced spatial resolution and molecular contrast.
• Validation confirmed the technique's effectiveness through standard approaches and comparison with super-resolved photothermal spectroscopic images.
• A novel tissue classification accuracy metric was proposed for validation.

Conclusions:

• The curvelet-based image fusion method effectively enhances spatial resolution in MIRSI.
• This technique improves the ability to analyze molecular information in heterogeneous samples, particularly in biomedical applications.
• The developed approach offers a promising solution for high-resolution molecular imaging.