Methods: Thirty oral leukoplakia and ten normal mucosa were collected in the research. All spectra were excited with 1W of 1064.1 nm radiation from NIR Nd:YAG laser. Based on the histological diagnoses, the regions were divided to the normal and premalignant group. In the same time, the premaligant group was subdivided to low-, intermediate- and high- grade group. The information of the subtracted spectra of compared group was gained by the characteristic band. Gaussian radial basis function support vector machines was used to classify spectra and establish the diagnostic models.
Results: Compared with normal tissue, low-grade hyperplasia had higher peak intensity in 852,1054,1667,2930 cm-1, demonstrated the increasing contents of protein, especially the Ganimalon, and lipids; intermediate-grade hyperplasia had higher peak intensity in 618,852 cm-1, demonstrated the increasing contents of protein, especially the phenylalanine and tyrosine; high-grade hyperplasia had higher peak intensity in 2083, 935,1261,1658 and 2934 cm-1, demonstrated the increasing contents of protein, DNA and lipid. The accuracies of comparison between low-grade and normal, intermediate-grade and normal, high-grade and normal were 52.5%, 0%, 71.88%, which elucidated the low modeling ability in these groups.
Conclusion: Combined with support vector machines, near infrared Raman spectroscopy could detect the biochemical variations in oral normal and premalignant tissues, but could not establish diagnostic model accurately, which need the further investigation.