Early colorectal cancer detection: a serum analysis platform combining SERS and machine learning

Area of Science:

• Biomedical Engineering
• Analytical Chemistry
• Cancer Research

Background:

• Colorectal cancer (CRC) presents significant global health challenges due to high incidence and mortality.
• Late-stage diagnosis of CRC often leads to poorer treatment outcomes, highlighting the need for sensitive early detection methods.

Purpose of the Study:

• To develop and validate a novel approach for early colorectal cancer (CRC) detection using Surface-Enhanced Raman Scattering (SERS) combined with a Principal Component Analysis-Dynamic Weighted Nearest Neighbor (PCA-DWNN) model.
• To assess the efficacy of a synthesized Gold Nanocluster (AuNC) substrate for SERS enhancement in serum samples from a colorectal cancer mouse model.

Main Methods:

• Establishment of a colorectal cancer (CRC) mouse model and collection of serum samples.
• Synthesis of a high-performance Gold Nanocluster (AuNC) substrate for Surface-Enhanced Raman Scattering (SERS) analysis.
• Development of a Principal Component Analysis-Dynamic Weighted Nearest Neighbor (PCA-DWNN) model for classifying SERS spectra of serum at different CRC stages.

Main Results:

• The synthesized AuNC substrate demonstrated high sensitivity, reproducibility, uniformity, and stability.
• The PCA-DWNN model achieved excellent classification accuracy (97.5%) and robustness in identifying complex SERS spectra.
• Analysis of SERS spectra revealed changes in serum biomolecules (proteins, lipids, amino acids, carbohydrates) correlating with CRC progression.

Conclusions:

• The combination of SERS technology and the PCA-DWNN model shows significant potential for the early detection of colorectal cancer (CRC).
• This approach offers a promising novel strategy for clinical diagnostics, enabling more accurate and sensitive identification of cancer stages.