Discrimination of Nasopharyngeal Carcinoma from Noncancerous Ex Vivo Tissue Using Reflectance Spectroscopy

Abstract

Reflectance spectroscopy is a low-cost, nondestructive, and noninvasive method for detection of neoplastic lesions of mucosal tissue. This study aims to evaluate the capability of reflectance spectroscopy system under white light (400–700 nm) with a multivariate statistical analysis for distinguishing nasopharyngeal carcinoma (NPC) from nasopharyngeal benign ex vivo tissues. High quality reflectance spectra were acquired from nasopharyngeal ex vivo tissues belonging to 18 noncancerous and 19 cancerous subjects, and the combination of principal component analysis-linear discriminant analysis (PCA-LDA) along with leave-one-spectrum-out cross-validation (LOOCV) diagnostic algorithm was subsequently employed to classify different types of tissue group, achieving a diagnostic sensitivity of 73.7% and a specificity of 72.2%. Furthermore, in order to distinguish NPC from nasopharyngeal benign ex vivo tissues based on reflectance spectra simply, spectral intensity ratios of oxyhemoglobin (R540/R576) were used as an indicator of the carcinogenesis associated transformation in the hemoglobin oxygenation. This tentative work demonstrated the potential of reflectance spectroscopy for NPC detection using ex vivo tissue and has significant experimental and clinical value for further in vivo NPC detection in the future.