Statistical classification based on SVM for Raman spectra discrimination of nasopharyngeal carcinoma cell

Raman spectroscopy(RS) has shown its advantages in detecting molecular changes associated with tissue pathology, which makes it possible to diagnose with optical methods non-invasively and real-time. It is very important to validate an existing classification model using different algorithms used in the discrimination of normal and tumor cells. In this work, three algorithms of SVM (Support Vector Machine) are used to validate LDA classification model of nasopharyngeal carcinoma (NPC) cell lines and nasopharyngeal normal cell line. All of these three SVM algorithms use the same data set as the same LDA model and achieve great sensitivity and specificity. Experimental results show that LDA classification model could be supported by different SVM algorithms and this demonstrates our classification model is reliable and may be helpful to the realization of RS to be one of diagnostic techniques of NPC.