Diagnosis of nasopharyngeal carcinoma from serum samples using hyperspectral imaging combined with a chemometric method.

Yan Wu Chu,Feng Chen,Yun Tang,Tong Chen,Yun Xin Yu,Hong Lin Jin,Lian Bo Guo,Yong Feng Lu,Xiao Yan Zeng

DOI: https://doi.org/10.1364/OE.26.028661

IF: 3.8

2018-01-01

Optics Express

Abstract:Diagnosing nasopharyngeal carcinoma (NPC) is a significant challenge because of the highly complex process. We proposed an approach to diagnose NPC serum using a combination of hyperspectral imaging and weight-based principal component analysis. Samples were prepared by pressing boric acid into pellets for use as the sera substrate. The sera, collected from 100 healthy volunteers and 60 NPC patients, was dripped onto the surface of the substrate for hyperspectral imaging. The characteristic spectral bands were selected based on the variable weight obtained from a support vector machine (SVM) model, using principal component analysis (PCA) to reduce the dimension in the extracted bands. Obtained results show that the accuracy rate, sensitivity, and specificity between the NPC sera and the sera of the healthy controls reached extremely high levels of 99.15%, 98.79%, and 99.36%, respectively. For the model's consistency evaluation, we found that the Kappa and area under the curve (AUC) of the receiver operating characteristic (ROC) curve were 0.99 and 0.98, respectively. These results suggest that the developed approach could serve as a noninvasive diagnostic and screening tool for highly accurate and consistent detection of NPC. Hence, a combination of hyperspectral imaging (HSI) and a weighted principal component analysis (WPCA)-SVM model represents a powerful and promising tool for NPC diagnosis. (C) 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement