Optical screening of nasopharyngeal cancer using Raman spectroscopy and support vector machine

In this study we present the application of Raman spectroscopy combined with support vector machine (SVM) for the characterization of Nasopharyngeal Cancer (NPC) in the human sera. In total, 14 pathological samples from confirmed NPC patients and 15 normal sera samples from healthy individuals have been collected and used in this study. Raman system with 785 nm laser has been used for recoding Raman spectra of both NPC and normal sera samples. Subtle intensity variations have been observed at certain Raman bands between both groups. Principal Component Analysis (PCA) has been used for features transformation and for highlighting spectral differences, while SVM is employed for discriminating cancerous non-cancerous samples. SVM has the ability to transforms linearly inseparable data to high dimensional space, by using mathematical function (kernel) where it may be linearly separable. An SVM model with different kernel functions i.e. polynomial kernel and Gaussian radial basis function (RBF) have been used for classification of diseased samples from the normal one. The performance of the model has been evaluated using k-fold (k = 5) cross validation method. Best performance efficiency has been achieved with RBF.