PLANCK: super-multiplex optical imaging without labeling

Area of Science:

• Biomedical Optics
• Chemical Imaging
• Artificial Intelligence in Imaging

Background:

• Label-based optical imaging faces limitations in multiplexing, cost, and complexity.
• Label-free optical imaging offers simplicity but lacks molecular specificity.
• Current molecular imaging techniques struggle with scalability and cost-effectiveness.

Purpose of the Study:

• To develop a label-free super-multiplex optical imaging technique.
• To demonstrate the correlation between vibrational spectroscopy and mass spectrometry imaging.
• To enable prediction of numerous molecular species from label-free vibrational images.

Main Methods:

• Paired vibrational spectroscopic imaging and mass spectrometry imaging were systematically studied.
• Supervised learning models were built based on latent space correlations.
• The developed technology, PLANCK, was demonstrated using infrared and Raman-based vibrational imaging.

Main Results:

• A strong correlation (>0.9) was discovered between vibrational spectroscopic imaging and mass spectrometry imaging latent spaces.
• Spatial distribution of over 100 molecular species was successfully predicted from label-free vibrational images.
• PLANCK demonstrated cost-effective and scalable molecular identification in diverse tissue systems, including live imaging.

Conclusions:

• Label-free super-multiplex optical imaging is achievable through AI-driven analysis of vibrational spectroscopic data.
• PLANCK provides a cost-effective and scalable solution for molecular imaging in research and translation.
• This AI-powered approach unlocks hidden molecular information from vibrational data for broad applications.