ScholarWorks@성균관대학교

초록

Rapid detection of multiple cancers offers remarkable advantages, particularly when the type of cancer is unknown, and enables the discovery of undiagnosed cancers through screening. Cancer development causes abnormal alterations in metabolites, many of which are excreted in the urine. Recognizing changes in urinary metabolites can reveal specific characteristics of various cancer types and aid in their diagnosis. Herein, we analyzed 218 urine samples from healthy individuals and patients with pancreatic, prostate, lung, and colorectal cancers using plasmonic materials to obtain label-free surface-enhanced Raman scattering signals. Applying artificial intelligence-based analytical methods, we achieved a diagnostic accuracy of >99% across all five groups. To identify urinary metabolic biomarkers, we compiled a library of known urinary metabolites and applied the multivariate curve resolution–alternating least squares method, a bilinear algorithm model, to relatively quantify the secretion levels of selected metabolites. This approach combines SERS signals from plasmonic materials with AI analysis and mathematical modeling to propose biomarkers. The technology is expected to provide a versatile platform not only for multicancer diagnosis but also for diagnosing various diseases, offering a promising tool for early detection and personalized medicine.

키워드