Rapid detection of nasopharyngeal cancer using Raman spectroscopy and multivariate statistical analysis

Optical spectroscopic techniques, including Raman spectroscopy, have shown promise for in vivo cancer diagnostics in a variety of organs. In this study, the potential use of a home-made Raman spectral system with a millimeter order excitation laser spot size combined with a multivariate statistical analysis for the rapid detection and discrimination of nasopharyngeal cancer from normal nasopharyngeal tissue was evaluated. Raman scattering signals were acquired from 16 normal and 32 nasopharyngeal carcinoma tissue samples. Linear discriminant analysis (LDA) based on principal component analysis (PCA) and partial least squares (PLS) were employed to generate diagnostic algorithms for the classification of different nasopharyngeal tissue types. Spectral differences in Raman spectra between the two types of tissues were revealed; the normalized intensities of Raman peaks at 1,001, 1,207 and 1,658 cm^-1 were more intense for nasopharyngeal carcinoma tissue compared to normal tissue, while Raman bands at 848, 936 and 1,446 cm^-1 were stronger in normal nasopharyngeal samples. The PCA-LDA algorithm together with leave-one-out cross validation yields a diagnostic sensitivity of 81% and a specificity of 87%, while the PLS method coupled with subwindow permutation analysis improves the diagnostic sensitivity and specificity to 85 and 88%, respectively. Therefore, Raman spectroscopy combined with PCA-LDA/PLS demonstrated good potential for improving the clinical diagnosis of nasopharyngeal cancers.