Classification of ENT tissue using near-infrared Raman spectroscopy and support vector machines

摘要

A recent developed pattern recognition algorithm, Support Vector Machines (SVM), was employed to classify nearinfrared Raman spectroscopy data collected from normal and cancerous ENT tissues. Three types of classifiers, linear, 3rd order polynomial, and radial basis function, were used. Highest diagnostic accuracy was obtained by 3rd order polynomial with a sensitivity of 91.86% and a specificity of 100%. The possibility to simplify SVM implementation was also explored by using principal component analysis (PCA) to extract significant principal components. It was found that the first five principal components as the data inputs were already sufficient to produce sensitivities of 100% and specificities of 100% for all these three classifiers. Combination PCA and linear discriminant analysis (LDA) to classify these ENT data was also performed and analysis results show that both methods, combination PCA & SVM and PCA & LDA yielded comparable performance.