On the Detection of Caries Lesions in Human Teeth using VIS/NIR-Spectroscopy
Ruohonen, Mika (2012)
Kuvaus
Opinnäytetyö kokotekstinä PDF-muodossa.
Tiivistelmä
Dental caries affects nearly 100% of the adult population in most countries. The current methods for diagnosing dental caries are able to detect caries only at a relatively advanced stage. Minimally invasive dentistry requires that caries is detected at an early stage of development, and that its status can be monitored frequently.
The objective of this study was to investigate whether diffuse reflectance near-infrared spectroscopy can be used to detect dental caries lesions that are advanced enough to be detected with manual inspection with fiber-optic illumination. Positive results would support the possibility of using reflectance near-infrared spectroscopy for detecting caries lesions at an early stage.
A total of 24 tooth samples were measured with two spectroscopes that together covered the wavelength range 200 - 1706 nm, using a general purpose transmission dip probe. Only the wavelength range 420 - 1000 nm was included in detailed analysis. Five different binary classification methods were used to classify the samples as either healthy or as carious. The performance of each classifier was evaluated with 4-fold cross-validation. One of the classification methods was a binary-classification support vector machine.
The results of this study suggest that diffuse reflectance near-infrared spectroscopy is able to improve the diagnostic accuracy of manual inspection with fiber-optic illumination, at least when the inspection is done by a novice. This claim is contingent on an assumption that all healthy sites of enamel have spectra that somewhat resemble each other, and partly on an assumption that all carious lesions on enamel show increased scattering in the near-infrared range. The reliability of these results is limited by the author's ability to diagnose caries lesions with the said manual method, and by the samples' ability to represent the variance among sites of healthy enamel and among caries lesions, though.
The objective of this study was to investigate whether diffuse reflectance near-infrared spectroscopy can be used to detect dental caries lesions that are advanced enough to be detected with manual inspection with fiber-optic illumination. Positive results would support the possibility of using reflectance near-infrared spectroscopy for detecting caries lesions at an early stage.
A total of 24 tooth samples were measured with two spectroscopes that together covered the wavelength range 200 - 1706 nm, using a general purpose transmission dip probe. Only the wavelength range 420 - 1000 nm was included in detailed analysis. Five different binary classification methods were used to classify the samples as either healthy or as carious. The performance of each classifier was evaluated with 4-fold cross-validation. One of the classification methods was a binary-classification support vector machine.
The results of this study suggest that diffuse reflectance near-infrared spectroscopy is able to improve the diagnostic accuracy of manual inspection with fiber-optic illumination, at least when the inspection is done by a novice. This claim is contingent on an assumption that all healthy sites of enamel have spectra that somewhat resemble each other, and partly on an assumption that all carious lesions on enamel show increased scattering in the near-infrared range. The reliability of these results is limited by the author's ability to diagnose caries lesions with the said manual method, and by the samples' ability to represent the variance among sites of healthy enamel and among caries lesions, though.