2.1.

Sample Preparation

Extracted molars and bicuspids extracted for orthodontic reasons were obtained from patients in the San Francisco Bay Area. Tooth collection was approved by the University of California, San Francisco Committee on Human Research. The teeth were cleaned and sterilized with gamma radiation and kept in a (0.1% thymol) solution to prevent fungal and bacterial growth. Thirty teeth with suspected hypomineralization and/or demineralization from caries were chosen based on visual inspection. Teeth were chosen by developmental defect location—lingual and facial defects were preferred to interproximal lesions, and defects on the cusp tips were preferred to lesions in the grooves and fissures. These criteria were intended to increase the probability of selecting teeth with fluorosis and not caries. Dean’s fluorosis index was used to obtain teeth with various degrees of enamel fluorosis.¹

All thirty teeth were mounted on $1.2\times 1.2\times 3{\mathrm{cm}}^3$ rectangular blocks of black orthodontic composite resin with the occlusal surface facing out from the square surface of the block. Each rectangular block fit precisely in an optomechanical assembly that could be positioned with micron accuracy. The rectangular symmetry of the mount facilitates matching the position of plano-parallel OCT scans to the thin sections produced for future mineral density determination using digital transverse microradiography (TMR).

2.2.

NIR Imaging

The NIR imaging setup is shown schematically in Fig. 1 . Light from a single-mode fiber-pigtail coupled to a $1310\text{-}\mathrm{nm}$ superluminescent diode (SLD) with an output power of $15\mathrm{mW}$ and a $35\text{-}\mathrm{nm}$ bandwidth, Model SLED1300D20A (Optospeed, Zurich, Switzerland), was used as the illuminating source. We found that broadband SLDs were advantageous to avoid speckle. An InGaAs focal plane array (FPA) $(318\times 252\text{pixels})$ the Alpha NIR (Indigo Systems, Goleta, California) with a Infinimite video lens (Infinity, Boulder, Colorado) was used to acquire all images. The acquired $12\text{-}\text{bit}$ digital images were analyzed using IR Vista software (Indigo Systems, Goleta, California). Two modes of imaging were used for imaging decay on proximal and occlusal surfaces that we call NIR-trans and NIR-occlusal mode imaging, as illustrated in Fig. 1.

Fig. 1

NIR imaging setup for NIR-trans (top) and NIR-occlusal (bottom) of tooth.

The NIR-trans mode of collecting NIR images is shown schematically in Fig. 1. In the NIR-trans mode images, the broadband light source and the InGaAs FPA are opposite each other, and crossed NIR polarizers, Model K46-252 (Edmund Scientific, Barrington, New Jersey), were inserted to remove light that directly illuminated the array without passing through the tooth.²²

For NIR-occlusal mode imaging, light enters the teeth just above the gumline and is highly scattered by the dentin. The diffusely scattered light migrates upward toward the surface of the crown, into the occlusal surface of the tooth. The enamel of the crown (outer white area) is transparent at $1310\mathrm{nm}$ and varies from $1\text{to}3\mathrm{mm}$ in thickness.²³

Reflected light images in the visible range were acquired of each occlusal surface using a color $1∕3\mathrm{in.}$ CCD camera with a resolution of 450 lines, Model DFK 5002/N (Imaging Source, Charlotte, North Carolina), equipped with the same Infinimite video lens.

2.3.

Polarization-Sensitive Optical Coherence Tomography

An all single-mode fiber autocorrelator-based optical coherence domain reflectometry (OCDR) system with polarization switching probe, high-efficiency piezoelectric fiber-stretchers, and an InGaAs receiver that was fabricated by Optiphase, Inc. (Van Nuys, California), was used to acquire images of the occlusal and smooth surface tomography of hypomineralized and demineralized teeth. This OCDR system was coupled with a broadband high-power superluminescent diode (SLD) from Denselight (Jessup, Maryland) with an output power of $10\mathrm{mW}$ and a bandwidth of $83\mathrm{nm}$ for a spatial resolution of $12\mu \mathrm{m}$ in air and $7.5\mu \mathrm{m}$ in enamel. A high-speed XY-scanning system with an ESP 300 controller and 850G-HS stages from National Instruments (Austin, Texas) was integrated with the SLD for in vitro optical tomography. The probe was designed to provide a spot diameter of $50\mu \mathrm{m}$ over a range of $10\mathrm{mm}$ . The system has been described in greater detail in Refs. 24, 25. The PS-OCT system was controlled using Labview software from National Instruments (Austin, Texas). The intensity of backscattered light was measured as a function of depth within the tissue. Two-dimensional (2-D) OCT intensity plots were obtained by laterally scanning the beam across the tooth.

2.4.

Digital Microradiography

A custom-built digital microradiography (TMR) system was used to measure the volume percent mineral content in the areas of hypomineralization on the tooth sections. High-resolution microradiographs were taken using Cu ${\mathrm{K}}_{\mathrm{a}}$ radiation from a Philips 3100 x-ray generator and a Photonics Science FDI x-ray digital imager (Microphotonics, Allentown, Pennsylvania). The x-ray digital imager consists of a $1392\times 1040\text{pixel}$ interline CCD directly bonded to a coherent micro fiber-optic coupler that transfers the light from an optimized gadolinium oxysulphide scintillator to the CCD sensor. The pixel resolution is $2.1\mu \mathrm{m}$ , and images can be acquired in real time at a frame rate of $10\mathrm{fps}$ . A high-speed motion control system with Newport UTM150 and 850G stages and an ESP 300 controller coupled to a video microscopy and laser targeting system was used for precise positioning of the tooth sample in the field of view of the imaging system. The experimental setup is shown in Ref. 12.

3.1.

NIR Imaging of Developmental Defects

Figures 2, 3, 4 show visible and NIR images of four teeth with developmental defects. In Fig. 2, two teeth are shown, comparing images taken in visible reflected light and NIR transillumination (NIR-trans mode of Fig. 1). The first tooth of Figs. 2a and 2b shows a large stained defect with the defect visible over a large portion of the tooth crown. In contrast, the NIR image shows the lesion over three more highly localized areas. There are differences between the visible and NIR images, and the NIR images show only the most severe areas of decay in the NIR-trans imaging mode. PS-OCT images of the same tooth along with images of a histological thin section are shown in Fig. 5 and indicate that the opaque areas in the NIR-trans image match the areas of deep penetration of the defect into the enamel while the areas of shallow penetration are not visible in the NIR-trans images. In the second tooth, Figs. 2c and 2d, the visible and NIR images are a better match, and both images show the defect over most of the upper part of the crown of that tooth.

Fig. 2

Visible reflected light images (a) and (c) and NIR-trans images (b) and (d) of two teeth covered by mild and moderate development defects. Note the “white caps” in (a) and (c), characteristic of teeth with development defects due to fluorosis.

Fig. 3

Visible reflected light images (a) and (b) and an NIR-occusal image (c) of a tooth covered by mild fluorosis and likely caries and stains in the occlusal grooves. Note that areas of mild fluorosis along with the stains are not apparent in the NIR-occlusal image, while the areas likely to be caries (white box) are visible with high contrast.

Fig. 4

Visible reflected light images (a) and (b) and an NIR-occusal image (c) of a tooth covered by mild fluorosis and either caries or more severe defects on the occlusal surface. Note that areas of mild fluorosis are not apparent in the NIR-occlusal image, while areas of more severe defects/caries are visible, as shown by the black arrows. The more severe areas of demineralization in the corresponding histological thin section (d) taken along the dotted line in (c) matches the more opaque areas of the NIR-occlusal image (white arrows).

Fig. 5

PS-OCT parallel (∥-axis) image (a) and orthogonal (⊥-axis) image (b) are shown along with the visible reflected light image (c) with a more serious defect. Visible (d) and near-IR (e) images of the histological thin section cut in the same position of the PS-OCT taken along the dotted line and arrow of (c) are also shown. (Color online only.)

The most interesting contrast between the visible light and NIR images is provided by the occlusal views. In Fig. 3, we have buccal and occlusal reflected visible light images of a molar with “white caps,” most likely due to hypomineralization. The tooth is also heavily stained on the occlusal surfaces. An NIR-occlusal image shows opacities along the fissures where lesions due to caries are likely to be found; the white caps, on the other hand, appear transparent in the NIR-occlusal image. Similar behavior is observed on other teeth, namely, the high contrast between sound and demineralized enamel in the NIR, while areas with suspected fluorosis have slightly increased contrast or cannot be seen in the NIR. Many of the developmental defects appear transparent, while obvious caries appear opaque in NIR light. In Fig. 4, we have a tooth in which the entire occlusal surface is completely covered by apparent hypomineralization, with white cusps or “snow caps” whiter than the surrounding sound enamel due to the increased porosity of the developmental defect. In the corresponding NIR image, Fig. 4c, most of the cusps still appear whiter than the surrounding sound enamel; however, there are two areas on the occlusal surface that appear more opaque than the surrounding enamel. The tooth was sectioned along the dotted line shown in Fig. 4c, and the section is shown in Fig. 4d. Most of the hypomineralization is localized to the enamel surface, with the exception of two large areas indicated by the two arrows where the decay or defect penetrates all the way through the enamel to the dentin. Those deeply penetrating areas of the defects/decay match the opacities in the NIR-occlusal images quite well. The differences in appearance appear to be related to the depth or severity of the developmental defect or caries lesion, the location on the tooth, and the imaging geometry.

3.2.

PS-OCT Scans of Developmental Defects

PS-OCT was used to scan suspected sites of developmental defects on several teeth. Namely, white areas on smooth facial surfaces and cusps where demineralization due to caries is seldom found were chosen. Images of suspected developmental defects on two teeth are shown in Figs. 5 and 6 , along with reflected light images and NIR-trans images of thin sections corresponding to the area scanned using the PS-OCT system. There are two images for each OCT scan. One corresponds to the light reflected in the same polarization state as the incident linearly polarized light (∥-axis), while the second image corresponds to the light reflected in the orthogonal or perpendicular polarization state (⊥-axis). Birefringence and depolarization from strong scattering such as that caused by demineralization in caries lesions or the porosity of developmental defects cause increased intensity in the perpendicular polarization state (⊥) versus the reflectivity in the original polarization (∥). Moreover, the strong reflection at the tooth surface that prevents resolution of the structure of the defect near the enamel surface does not depolarize the incident light, so it is easier to resolve the structural differences in the defect in the orthogonal polarization image, the ⊥-axis image. The red-dashed line across the smooth surface of the tooth shows the scan direction. The reflected intensity reading for each point (measured in dB), is shown from black to red to yellow, with yellow as the most intense reflectivity. The PS-OCT scan in Fig. 6 shows a shallow defect localized to the outer layer of enamel. This developmental defect is fairly uniform across the entire tooth surface, in contrast to a caries lesion, which is typically more localized. The areas of greater intensity in the PS-OCT images match the areas of the defect shown in the visible and NIR images of the histological thin sections of Figs. 6d and 6e very well. The ⊥-axis PS-OCT image, Fig. 6b, shows a thin transparent zone just below the enamel surface. This zone is indicative of a zone of higher mineral content above the body of the defect similar to what is observed after remineralization of caries lesions. That zone is also very apparent in the NIR-trans image of the thin section while it is very difficult to resolve in the visible light image of the thin section, Fig. 6d or the ∥-axis PS-OCT image, Fig. 6a. Similar behavior is observed for the more severe defect shown in Fig. 5. The PS-OCT images match the position, depth, and severity of the defects shown in the thin sections quite well, indicating that PS-OCT can be used to measure the depth and severity of the defects due to the increased light scattering from the increased porosity. A surface zone is also found in the more severe defect shown in Fig. 5, and that zone is clearly resolved only in the ⊥-axis PS-OCT and the NIR-trans images.

Fig. 6

PS-OCT parallel (∥-axis) image (a) and orthogonal (⊥-axis) image (b) are shown along with the visible reflected light image (c) of a tooth with a shallow defect. More intense areas of reflectivity are yellow and red in the false-color PS-OCT images. Visible (d) and near-IR (e)images of the histological thin section cut in the same position of the PS-OCT taken along the dotted line and arrow of (c) are also shown. The defect appears whiter in the visible (d) and darker in the near-IR (e) images of the thin section; it is located on the top of the section and runs from the dentinal enamel junction to the cusp. (Color online only.)

3.3.

Optical Properties of Developmental Defects

In order to determine whether there are fundamental differences in the nature of light scattering between hypomineralization due to developmental defects such as fluorosis and demineralization by caries, the optical attenuation of suspected defects of varying severity or mineral content was measured at $1310\mathrm{nm}$ in the near-IR. Near-IR images and x-ray microradiographs were acquired of thin tooth sections with mild to moderate developmental defects for comparison. The attenuation coefficients or scattering coefficients were acquired from the NIR images by measuring the attenuation of NIR light through $200\text{-}\mu \mathrm{m}$ -thick sections cut through the developmental defects. Absorption at this wavelength was assumed to be negligible. Each NIR image was processed by converting from relative intensity values to values of attenuation $\left({\mu }_{\tau }\right)$ in units of inverse centimeters. The volume percent mineral content of each microradiography image was determined from a calibration curve of x-ray intensity versus the sample thickness of sound enamel sections. Line profiles through defect areas on 10 samples were selected, and 10 points from each sample of varying mineral content were plotted on a graph. A plot of the optical attenuation versus volume percent mineral loss was assembled from all the line profiles and is shown in Fig. 7 . Points from each independent sample manifest similar behavior with mineral loss. Nonlinear regression using an exponential growth curve shows that the optical scattering increases exponentially with mineral loss and that the magnitude of light scattering approaches a maximum after 10% mineral loss.

Fig. 7

A plot of the optical attenuation $\left({\mathrm{cm}}^{-1}\right)$ of 100 points taken from 10 tooth sections ( $200\text{-}\mu \mathrm{m}$ thick) with developmental defects of varying severity (10 points per sample). The solid line represents the best exponential fit $r^2=0.75$ to the attenuation coefficients versus volume percent.