ИСТИНА

Label-free fluorescence spectroscopy is a powerful tool for biomedical diagnostics. This approach is based on the signal from endogenous fluorophores such as NADH and flavins (metabolic imaging), bilirubin (liver disease), lipofuscin (oxidative processes), melanin (melanoma) etc. However, a number of organs and structures exhibit fluorescence signal, whose origin is unknown – e.g., atherosclerotic plaques, parathyroid gland, amyloids and some tumours. Here we will discuss a framework for prediction of optical properties of molecules in the human organism with machine learning approach and demonstrate how it helps in understanding fluorescence formation. The examples of clinical translation of fluorescence spectroscopy will be also presented, including NIR autofluorescence.