Visualization and quantification of healthy and carious dentin structure using confocal laser scanning microscopy

In this study, a fluorescence technique was developed for visualization of dentin using confocal laser scanning microscopy (CLSM). Eighteen extracted human teeth were used: 13 showing no clinical signs of caries and 5 with visually apparent decay. Preliminary study: All teeth were horizontally sectioned to approx. 200 micrometers thickness and pre-treated as follows: no pretreatment; vacuum only; ultrasonication only; sodium hypochlorite (NaOCl) only; vacuum and NaOCl; ultrasonication and NaOCl; or vacuum, ultrasonication and NaOCl. Samples were stained with Rhodamine 123 fluorescent dye at a concentration of 10-5 M in phosphate buffer saline for 1 to 24 hours. Caries study: Dentin surfaces, some with pre-existing caries, were visualized using CLSM. Most dentin tubules in sound dentin appeared open using CLSM, but most dentin tubules in carious dentin appeared closed or narrowed. Surface images obtained using CLSM were similar to those seen by SEM, but additional subsurface imaging was possible using CLSM at depth intervals of 1 micrometers to a depth of 30 - 50 micrometers . This technique shows good potential for non-invasive surface and subsurface imaging of dentin structures.

for assessment of dental operative procedures and restorations,36 study of the tooth/restoration interface,7 evaluation of enamel cutting effects,8'9 visualization of enamel surface changes,'° and clinical dental research.1' However, this technique has not been used for visualization and quantification of dentin, of the caries process, and more specifically of carious dentin.
CLSM systems can scan a laser beam over stationary samples.'2 Surface images obtained using the confocal microscope are similar in character to those provided by scanning electron microscope (SEM), but subsurface imaging is also possible. Furthermore, confocal microscope images can be three dimensionally reconstructed.'2"3 Conventional dehydration and gold-coating methods utilized for SEM have several disadvantages when compared to sample preparation for CLSM: dehydration and heating can damage specimens and cause a variety of artifacts such as cracking or 'bubbling". Tooth samples can be visualized using the confocal laser scanning microscope either by lightly metal-coating the surface, or by using fluorescence techniques.
Two basic types of confocal microscopy exist: real-time direct view tandem scanning microscopy (TSM) and CLSM.'2 The high-frame speed of the TSM allows real-time examination of teeth in vivo.14'15 In contrast, all CLSM's rely on powerful image processing devices and frame stocks for information displays.
The acquisition of perfectly registered, serial optical sections by this technique provides an excellent means for the three dimensional reconstruction and quantification of objects.
The purpose of this study was to develop a technique for use of CLSM for imaging and study of dentin. In addition to identifying appropriate dyes and fluorescence staining techniques, imaging depth and resolution were investigated. Results were compared with images from conventional SEM techniques to contrast the levels of information obtained and to assess any structural effects of the staining techniques developed in this study. Development and application of appropriate techniques for CLSM imaging and study of healthy and carious dentin was also priority. Moreover, the sensitivity of the system for imaging natural and induced caries conditions was investigated.

Sample preparation
Seven extracted human teeth showing no clinical signs of caries, stored in demineralized water with 0.01 % (w/v) thymol, were horizontally sectioned into thin slices (approx. 200 .tm thickness) using a low speed saw with coolant (Isomet, Buehler, IL, U.S.A.).

Pretreatment study
To determine optimal pretreatment conditions and procedures we undertook the following preliminary SP!E Vol. 2672 165 investigations. Samples were treated with NaOC1 (5.25 % by Wt, Darrow Comp., CA, U.S.A.) solution, vacuum and/or ultrasonication for 1 h. After pretreatment, samples were subjected to staining procedures.
Rhodamine 123 (Eastman Kodak CO., NY, U.S.A.) with an absorption peak at 51 1 nm was used as the fluorescent dye, as this absorption peak corresponds most closely to the excitation wavelength of 488 nm used in our confocal laser microscope. Staining was performed at the concentration of iO M for 3 h. After staining, samples were rinsed 2-3 times with demineralized water for 5 minutes. Sections were fixed to slide glasses with cyanoacrylate glue, and observed through the CLSM. Later, the same samples were prepared for conventional SEM procedures.

Staining study
To identify optimal fluorescence staining dyes and procedures we undertook the following preliminary investigations. In untreated samples, Rhodamine 123 concentrations of iO, iO, iO, i06, and iO M were tested at pHs 6.0, 6. Germany). The laser wavelength of 488 nm was used for fluorescence excitation; emission was isolated with a long pass 520 nm filter. The distance between optical sections was 1 .tm or 2 .tm on the Z-axis.
Overall depth of acquisition ranged from approx. 20 im to 60 im depending on the depth of penetration of Rhodamine 123 into the sample. The information obtained was stored on 1 GByte optical disc (Panasonic) and 3-dimensional images were generated from stacks of stored images using original LSM 410 software. 66 ISPIE Vol. 2672

Sample preparation
Six extracted human teeth showing no clinical signs of caries and decay were selected. Three of these samples constituted group A (sound teeth). In the remaining 3 samples, caries-like lesions were artificially induced as described below (group B). Five additional teeth with pre-existing naturally occurring dentin caries made up group C of this study. Following factors were set as inclusionary criteria for the "natural caries" group: dentin was used from the tooth crown just below the amelodentinal junction and sections were prepared parallel to the amelodentinal junction; no gross structural cavitation or destruction as determined by the naked eye was apparent in the samples used; only darkly discolored carious dentin was used; and natural caries extended through the entire thickness of the sections under investigation. These criteria were established in an attempt to set a minimal standard of comparability between carious specimens. All teeth were stored in demineralized water with 0.01 % (w/v) thymol, and were horizontally Comparison between SEM and CLSM The images obtained by SEM or CLSM were very similar at a magnification of up to l000x. This is the magnification limit of CLSM, but additional subsurface imaging is also possible.

Caries study
CLSM results in teeth with naturally-occurring caries fell into one of 3 groups: The first type; Many dentin tubules in the carious lesion were closed or extensively narrowed, usually at a subsurface level. Thus, on the surface of the carious lesion, tubules are visible, but at deeper levels many could no longer be visualized. Dye penetration into the carious area was evidently very poor compared to the sound zone.
The second type; Here, typically, some dentin tubules in the carious area were narrowed, or completely blocked. The margins of the dentin tubules affected by caries were stained strongly compared to tubules in sound dentin. Dye penetration in these samples was reduced compared to healthy samples with visibility becoming problematic already at a depth of 15 im or less.
The third type; Some dentin tubules were narrowed, only a few were completely blocked, and the margins of dentin tubules were again strongly stained. The dark zones seen in the CLSM images corresponded to patches of extensive carious destruction of the dentin surface.
Artificial caries-like lesions were investigated after 3 days induction. Many dentin tubules in the artificial caries-like lesion were still open, but dye penetration into the carious lesion was reduced as compared to sound dentin. The margins of the dentin tubules were strongly stained compared to sound dentin. The etched surface appearance was attributed to caries induction methods causing dentin surface dissolution.

DISCUSSION
A limited range of applications for scanning confocal microscopy has been investigated in the field of dentistry. Usually these involve methods for assessment of dental operative procedures and rial31 By eliminating the need for problematic and tedious specimen preparation techniques, confocal microscopy is well suited to the observation of dental and material surfaces, and to monitoring effects of various agents and factors on their microstructure. In these previous studies, however, TSM was mainly used, providing poor horizontal resolution.'2 Moreover, this technique has not been used for visualization and quantification of caries processes, and more specifically of carious dentine. Kodaka et al. reported dental applications of CLSM, but using metal-coated samples rather than fluorescent staining technique.'° In addition to its costly, practically demanding and time-consuming nature, metal coating has the disadvantage of usually requiring sample dehydration, a process which can easily damage, alter or distort sample structure and produce multiple artifacts.
SP!E Vol. 2672 / 69 Compared with SEM, magnification provided by confocal microscopy and fluorescence techniques is low (maximum xl000), but within this range, CLSM allows surface and sub-surface visualization in three dimensions. Dimensional quantification is easy and accurate: thus we were able to measure the depth of carious lesions easily using marker systems on the computer screen. Disadvantages of CLSM techniques include its lack of suitability for clinical investigations, as relatively thin sections are needed. However, the high-frame speed of the TSM enables real-time examination of teeth in vivo.14'15 Our investigations of fluorescence staining techniques for dentin showed that NaOC1 application only did not enhance dye uptake, but was very effective together with ultrasonication in achieving good staining.
Similar surface images were obtained using SEM and CLSM, but it was possible to perform also noninvasive subsurface imaging in dentin using CLSM. The presence of a smear layer did reduce the effectiveness of fluorescence staining, but the smear layer was partially permeable to the fluorescent dye.
Using the fluorescence techniques developed in this investigation, we were able to achieve good surface and subsurface images of dentin samples. Optimum staining conditions for visualization and quantification of dentinal structure were identified in this study. However, further improvements in subsurface visualization would be beneficial.
In the second part of this investigation, decayed dentin samples without obvious gross carious destruction or cavitation were used. Caries-induced zones of dentin destruction, demineralization and tubular occlusion were clearly visible using CLSM in our "natural caries" samples. These findings correspond with histopathological observations described by other authors.17'18 Demineralization of the dentinal tubular walls and their immediate surroundings was also observed in some carious dentin samples. This observation may be attributed to the caries process. However, we noticed a similar effect in some healthy dentin tissues, after sample preparation using NaOCI and ultrasonication to enhance dye uptake. The same sample preparation technique with NaOC1 and ultrasonication was used in these specimens. Thus, at this stage the exact origins of this observation remain to be elucidated, and further studies are necessary to assess optimal sample preparation techniques for CLSM.
The difference in appearance of "natural" vs "induced" caries in the CLSM was sinking. Very few occluded or narrowed dentin tubules were observed in the "artificial caries" samples, and the entire sample surface had a strongly demineralised appearance. These results are in agreement with histopathological observations by others,'7"8 arid are mainly attributed to the differences between the chemical induction techniques used in artificial caries, and the bacterially-related processes and pulpal reactive events occurring during natural decay.
We used CLSM for visualization and quantification of caries process and more specifically of caries dentin, and obtained useful and novel results. However, this technique requires development to optimize its potential. 70 / SPIE Vol. 2672