Metabolic Map of Colorectal Cancer and Artificial Intelligence‐Aided Identification Based on Single‐Cell Raman Spectroscopy and Metabolomics

ABSTRACT

Colorectal cancer (CRC) remains a leading cause of cancer‐related mortality, underscoring the need for improved early detection. Single‐cell Raman spectroscopy (SCRS) enables the detection of CRC‐associated biomolecular alterations and shows promise as a label‐free and complementary auxiliary tool for improved CRC screening and characterisation. This study employed confocal Raman spectroscopy and imaging to analyse CRC cells and normal cell lines alongside 14 paired tissue samples. Biomolecules were quantified from the SCRS data to construct a machine learning classifier, and ultra‐performance liquid chromatography–tandem mass spectrometry (UPLC–MS/MS) was used for metabolomic profiling. Metabolic mapping revealed consistent alterations in lipids, nucleic acids, proteins, amino acids and sugars across CRC and normal cell lines using Raman imaging and UPLC–MS/MS. Lipids were downregulated in CRC, whereas reduced sugar levels correlated with the tumour differentiation potential. The classification model exhibited high efficacy in diagnosing CRC and predicting its T stage and N stage, with AUC values ranging from 0.94 to 1.00. This study revealed the critical roles of lipid and sugar metabolism in CRC development and differentiation, establishing a foundation for understanding the pathogenesis and identifying therapeutic targets. The AI‐assisted SCRS analysis showed high accuracy in classifying cancer, indicating its promise as a simple, label‐free auxiliary tool for CRC screening and diagnosis.