Early detection of dental fluorosis using Raman spectroscopy and principal component analysis

Raman spectroscopic technique has the potential to provide vibrational spectra of minerals by analyzing scattered light caused by monochromatic laser excitation. In this paper, recent applications of Raman spectroscopy in the study of dental hard tissues are reported. Special attention is given to mineral components in enamel and to calcium fluoride formed in/on enamel. The criteria used to classify the dental hard samples were according to the Dean Index (DI), which consists into healthy or control, mild, moderate, and severe, indicating the amount of dental fluorosis observed on enamel. A total of 39 dental samples (9 control, 9 mild, 10 moderate, and 11 severe) were analyzed in the study. Dental samples were positioned under an Olympus microscope, and around 10 points were chosen for Raman measurement. All spectra were collected by a Horiba Jobin-Yvon LabRAM HR800 Raman Spectrometer with a laser of 830-nm and 17-mW power irradiation. Raw spectra were processed by carrying out baseline correction, smoothing, and normalization to remove noise, florescence, and shot noise and then analyzed using principal component analysis (PCA). In the spectra of dental samples, we observed the main bands as the broad band due to CO 3 2 − (240–300 cm −1 ), CaF 2 (322 cm −1 ), PO 4 3 − ν 2 vibrations (437 and 450 cm −1 ), PO 4 3 − ν 4 vibrations (582, 598, and 609 cm −1 ), PO 4 3 − ν 1 vibrations (960 cm −1 ), PO 4 3 − ν 3 vibrations (1,045 cm −1 ), and CO 3 2 − ν 1 vibration (1,073 cm −1 ). Nevertheless, the intensity of the band at 960 cm −1 associated to symmetric stretch of phosphate, PO 4 3 − ν 1 , decreases as the amount of dental fluorosis increases, suggesting that the intensity of this band could be used to quantitatively measure the level of fluorosis on a dental sample. On the other hand, PCA allowed to identify two large clusters discriminating between control, and severe and moderate samples with high sensitivity and specificity. PCA was able to discriminate mild from moderate samples with 100 % sensitivity and 89 % specificity and mild from severe samples with 91 % sensitivity and 100 % specificity. In addition, PCA was also able to discriminate between mild samples and group formed by the moderate and severe samples with 95 % sensitivity and 89 % specificity. Finally, PCA allowed us to define the wavelength differences between the spectral bands of the healthy teeth with sound enamel and those with fluorosis by confirming that the main chemical differences