Hyperspectral Raman Imaging for Automated Recognition of Human Renal Amyloid

Area of Science:

• Biomedical Engineering
• Analytical Chemistry
• Pathology

Background:

• Current identification of amyloid deposits relies on histochemical staining for common types (AL, AA) and mass spectrometry for rarer forms.
• Raman spectroscopic imaging offers a label-free method for chemical mapping and molecular characterization of tissues.
• Developing advanced diagnostic tools is crucial for accurate amyloidosis subtyping and patient management.

Purpose of the Study:

• To assess the feasibility of using Raman spectroscopy with artificial intelligence (AI) for detecting and characterizing amyloid deposits in kidney biopsies.
• To differentiate between light-chain amyloid (AL), serum amyloid A (AA), and non-amyloid (NA) tissues using this novel approach.
• To establish a proof-of-concept for a non-invasive, rapid diagnostic method for renal amyloidosis.

Main Methods:

• Acquisition of Raman hyperspectral images from unstained frozen kidney tissue sections.
• Application of three distinct machine learning-assisted analysis models to the hyperspectral image data.
• Testing the models on biopsies with confirmed AL amyloidosis (λ and κ types), AA amyloidosis, and no amyloidosis (NA).

Main Results:

• The AI-assisted Raman spectroscopy models achieved high accuracy, distinguishing between AL, AA, and NA tissues in 93-100% of cases.
• The technique successfully characterized the molecular composition of amyloid deposits in the tested kidney biopsies.
• This proof-of-concept study demonstrated the potential for precise subtyping of renal amyloidosis.

Conclusions:

• Raman spectroscopy coupled with AI presents a promising, label-free method for identifying and subtyping renal amyloidosis.
• This approach could potentially streamline the diagnostic workflow for amyloidosis, reducing reliance on traditional staining methods.
• Further validation is warranted, but these preliminary findings highlight a significant advancement in amyloidosis diagnostics.