Extracting autofluorescence spectral features for diagnosis of nasopharyngeal carcinoma

Abstract

The aim of this study is to investigate the autofluorescence spectral characteristics of normal and cancerous nasopharyngeal tissues and to extract the potential spectral features for diagnosis of nasopharyngeal carcinoma (NPC). The autofluorescence excitation-emission matrix (EEM) of 37 normal and 34 cancerous nasopharyngeal tissues were recorded by a FLS920 spectrofluorimeter system in vitro. Based on the alteration in proportions of collagen and NAD(P)H, the integrated fluorescence intensity of I455 ± 10 nm and I380 ± 10 nm were used to calculated the ratio values by a two-peak ratio algorithm to diagnose NPC tissues at 340 nm excited. Furthermore by applying the receiver operating characteristic curve (ROC), the 340 nm excitation yielded an average sensitivity and specificity of 88.2 and 91.9%, respectively. These results may have practical implications for diagnosis of NPC.