Near-infrared confocal micro-Raman spectroscopy combined with PCA–LDA multivariate analysis for detection of esophageal cancer

Longping Chen,Yue Wang,N. Liu,Duo Lin,Cuncheng Weng,Jixue Zhang,Li-Huan Zhu,Weisheng Chen,Rong Chen,S. Feng

DOI: https://doi.org/10.1088/1054-660X/23/6/065601

2013-04-30

Abstract:The diagnostic capability of using tissue intrinsic micro-Raman signals to obtain biochemical information from human esophageal tissue is presented in this paper. Near-infrared micro-Raman spectroscopy combined with multivariate analysis was applied for discrimination of esophageal cancer tissue from normal tissue samples. Micro-Raman spectroscopy measurements were performed on 54 esophageal cancer tissues and 55 normal tissues in the 400–1750 cm−1 range. The mean Raman spectra showed significant differences between the two groups. Tentative assignments of the Raman bands in the measured tissue spectra suggested some changes in protein structure, a decrease in the relative amount of lactose, and increases in the percentages of tryptophan, collagen and phenylalanine content in esophageal cancer tissue as compared to those of a normal subject. The diagnostic algorithms based on principal component analysis (PCA) and linear discriminate analysis (LDA) achieved a diagnostic sensitivity of 87.0% and specificity of 70.9% for separating cancer from normal esophageal tissue samples. The result demonstrated that near-infrared micro-Raman spectroscopy combined with PCA–LDA analysis could be an effective and sensitive tool for identification of esophageal cancer.

Physics