Combined classifier for discriminating cancerous tissue from normal tissue using light-induced autofluorescence

Abstract

We investigated a novel method combining principal component analysis (PCA) and supervised learning technique, support vector machine (SVM), for classifying carcinoma lesion from normal tissue with light-induced autofluorescence. The autofluorescence spectral signals were collected in vivo from 85 nasopharyngeal carcinoma lesions and 131 normal tissue sites from 59 subjects during routine nasal endoscopy. With the combined PCA and SVM classifying algorithm, the achieved overall accuracy is over 97%, companied with 95% sensitivity and 99% specificity for discriminating carcinoma from normal tissue. In comparison with the previously developed algorithms based on PCA method, this new method outperforms threshold- and probability-based PCA algorithms in all instances. The experimental results indicate great promise for autofluorescence spectroscopy based detection of small carcinoma lesion in the nasopharynx and other tissues.