Detection of aging-induced vascular remodeling based on Raman imaging and deep learning

Vascular aging-related remodeling is a common pathological basis for many chronic diseases, so early detection of physical arterial aging is important for their prevention and control. Existing staining methods can only analyze a limited number of tissue components at a time and often suffer from inaccuracies caused by over- or under-staining. In this study, we performed high-quality Raman imaging to simultaneously analyze five components in mouse aortic sections: elastic fibers, types I and III collagen fibers, nuclei, and cytoplasm of vascular smooth muscle cells (VSMCs), detailing their content and distribution changes. Despite subtle differences in Raman spectra, young and aged aortic tissues were successfully distinguished using multivariate curve resolution-alternating least squares (MCR-ALS) analysis and deep learning, achieving an AUC of 0.986 (95 % CI: 0.979-0.992). Additionally, Raman imaging and metabolomics revealed metabolic changes in arterial aging related to collagen synthesis and post-modifications, offering new potential therapeutic targets. Thus we show that Raman imaging combined with advanced algorithms is potentially useful in detecting vascular-aging remodeling, as well as monitoring the aging process.