Diagnosis of Occlusal Tooth Wear Using 3D Imaging of Optical Coherence Tomography Ex Vivo

The aim of this study was to assess the utility of 3D imaging of optical coherence tomography (OCT) for the diagnosis of occlusal tooth wear ex vivo. Sixty-three extracted human molars with or without visible tooth wear were collected to take digital intraoral radiography and 3D OCT images. The degree of tooth wear was evaluated by 12 examiners and scored using 4-rank scale: 1—slight enamel wear; 2—distinct enamel wear; 3—tooth wear with slight dentin exposure; 4—tooth wear with distinct involvement of dentin. The degree of tooth wear was validated by the histological view of confocal laser scanning microscopy (CLSM). The sensitivity, specificity, and area under the curve (AUC) of receiver operating characteristic analysis were calculated. Diagnostic accuracy was compared with the agreement with CLSM observation using weighted kappa. The results were statistically analyzed at a significance level of α = 0.05. Three-dimensional OCT showed significantly higher sensitivity (p < 0.05) for all the diagnostic thresholds of enamel wear and dentin exposure than digital radiography (0.82, 0.85, and 0.79 vs. 0.56, 0.52, and 0.57, respectively). Three-dimensional OCT showed higher AUC and kappa coefficients than digital radiography (p < 0.05), where mean AUC and Kappa values were 0.95 and 0.76 for OCT and 0.92 and 0.47 for radiography, respectively. No significant difference of specificity was observed (p > 0.05). Three-dimensional OCT could visualize and estimate the degree of tooth wear and detect the dentin exposure at the tooth wear surface accurately and reproducibly. Consequently, a new guideline for tooth wear assessment can be proposed using OCT.


Introduction
Tooth wear is defined as the loss of dental hard tissue by physical or chemical factors that occur throughout life [1]. The progression of tooth wear is irreversible and its prevalence increases with age [2,3]. The literature considers tooth wear resulting from a multifactorial etiology with interactions of physical and chemical agents. Although the changes resulting from physiological tooth wear are usually subtle and asymptomatic, excessive physical or chemical stimulation can lead without evidence of tooth wear at occlusal surfaces were collected. The teeth that had visible caries, restoration, or distinct cracks on the occlusal surface were excluded for this study.
The selected teeth were cleaned using a hand scalar and brush cone attached to a low-speed handpiece with a prophylaxis paste. After the removal of calculus or debris, the area of interest for the evaluation of tooth wear level was selected from the occlusal surface for each tooth. Digital photographs of occlusal surface, including the area of interest, were obtained with a digital camera (Nikon D50, Nikon, Tokyo, Japan) with an AF-S Micro Nikkor 105 mm lens with a fixed magnification of ×1.

Digital Dental Radiography
Digital dental radiographs of each tooth were captured with the X-ray tube positioned toward the buccal surface and horizontal aspect of the occlusal surface. A digital intraoral sensor was placed on the lingual aspect of the tooth, and an X-ray unit (Dentnavi Hands, Yoshida Dental, Tokyo, Japan) was operated at exposure settings of 60 kV and 7 mA, with an average exposure time of 0.63 s. Images were viewed on the monitor using the associated image program software (ClinicalView ver. 10.1, Instrumentarium Kavo Dental).

SS-OCT System
After taking digital dental radiographs, SS-OCT scanning was projected onto the occlusal surface including the area of interest to construct 3D images. The prototype SS-OCT system (Yoshida Dental, Tokyo, Japan) used in this experiment employed a frequency domain technique that measured the magnitude and time delay of reflected light to construct a depth profile. The wavelength ranged from 1240 to 1380 nm, with a central wavelength of 1310 nm and a sweep rate of 50 kHz. The high-speed frequency-swept laser was projected onto the occlusal surface using a hand-held scanning probe at a fixed distance to take 3D images. The power of the object beam was 18 mW and the system had a sensitivity of 100 dB. The optical resolution of the 3D dataset in air was less than 11 µm in depth and 40 µm in lateral and axial dimensions.

Scoring by Examiners
Twelve dentists were recruited (8,9,10,11, and 12, with less than 5 years of clinical experience; 6 and 7, with 5-10 years of clinical experience; 3, 4, and 5, with 11-20 years of clinical experience; 1 and 2, with over 21 years of clinical experience) and participated in this experiment as examiners. In order to reach a consensus on the evaluation criteria, the reference examiner (MK) discussed the radiographs and OCT images with 12 dentists in a 1 h session. For the calibration session, MK used 12 extracted teeth images that were not included in the study.
After the discussion, the examiners performed scoring of the tooth wear level of the occlusal surface independently using the following 4-rank scale: Score 1: Slight enamel wear. Initial tooth wear was within the enamel and more than 1/2 thickness of enamel left. Score 2: Distinct enamel wear. Tooth wear was within the enamel and less than 1/2 thickness of enamel left. Score 3: Tooth wear with slight dentin exposure. Tooth wear reached to the dentin-enamel junction (DEJ). Dentin exposure was slight with less than 1 mm diameter. Score 4: Tooth wear with involvement of dentin. Tooth wear was beyond the DEJ to cause dentin exposure. The diameter of dentin exposure was more than 1 mm.
A liquid crystal display monitor was used to display either digital radiographs or 3D OCT images, associated with the occlusal view of digital photographs. For the radiographs, the original image without enhancement of contrast or brightness was used. For OCT, 3D images of occlusal surface as well as the sequence of two-dimensional (2D) tomographic images extracted from the 3D dataset were dynamically displayed in video format using a custom-developed software (KakumaViewer, Yoshida Dental). Display settings such as brightness and contrast were unchanged from the default for all images and examiners.

Confocal Laser Scanning Microscope Observation
In order to validate the level of occlusal tooth wear, histological sectioning was done to observe the occlusal tooth surface directly under confocal laser scanning microscopy (CLSM) (VK-X150 series, Keyence, Osaka, Japan). The desired cross-sectional slides at the area of interest were marked on the teeth and were sectioned and trimmed from the buccal surface along the mesiodistal plane parallel to the tooth axis using a rotary diamond instrument attached to a highspeed handpiece under copious water coolant. The sectioned teeth were further trimmed using 2000-grit silicon carbide paper, followed by polishing with diamond paste down to 1 µm. The cross-sectioned polished surface was ultrasonicated with distilled water for 1 min to remove the polishing debris and was examined with CLSM at 100× magnification.

Statistical Analysis
The statistical analysis was performed using SPSS for Windows version 23 software (IBM, Armonk, NY, USA). In this study, slight enamel wear for score 1 was considered as physiological tooth wear (intact), whereas tooth wear for scores 2, 3 and 4 was considered as at the pathological level. The sensitivity and specificity of 3D OCT and digital radiography for the diagnosis of distinct enamel wear and involvement of dentin exposure were calculated for each examiner by cross tabulation. Sensitivity and specificity can be put into equations as below: In this study, sensitivity represents the probability of a diagnostic method identifying the tooth wear at a pathological level which in fact has pathological tooth wear. The higher the value of sensitivity, the more correctly the method can detect the pathological tooth wear.
Specificity represents the probability of a diagnostic method identifying the tooth wear at a physiological level which in fact has physiological tooth wear. The higher the value of specificity, the more correctly the method can detect the physiological tooth wear.
Area under the curve (AUC) of receiver operating characteristic (ROC) analysis was also measured using the results of sensitivity and specificity. The ROC curve was created by plotting the true positive (sensitivity) against the false positive rate (1-specificity). Consequently, the ROC curve depicts the relative trade-offs between true positive and false positive. The best possible diagnostic method is 100% sensitivity and 100% specificity with perfect classification yielding a point in the upper left corner. High AUC value represents good classification results of the method.
The diagnostic accuracy for both imaging methods was calculated using the agreement with histological CLSM observation by weighted Kappa. The results were statistically analyzed using the Wilcoxon rank sum test at a significance level of α = 0.05.

Results
In this study, a total of 63 molars with enamel wear at the occlusal surface were used. We validated the occlusal surfaces by CLSM observation of histological sections at magnification of ×100. As a result, the 63 teeth included 16 teeth with slight enamel wear (score 1), 18 teeth with distinct enamel wear (score 2), 13 teeth with tooth wear with slight dentin exposure (score 3), and 16 teeth with tooth wear with distinct involvement of dentin exposure (score 4).
Representative images of 3D OCT and digital radiography are shown in Figures 1-7. The sensitivity, specificity, and AUC values for the diagnosis of distinct tooth wear and involvement of dentin in each diagnostic threshold are shown in Tables 1 and 2. Three-dimensional OCT showed significantly higher sensitivity than the digital radiography for all the diagnostic thresholds of tooth wear level. The sensitivity of 3D OCT for distinct enamel wear, tooth wear with slight dentin exposure, and tooth wear involving dentin was 0.82, 0.85, and 0.79, respectively, whereas for digital dental radiography, it was 0.56, 0.52, and 0.57, respectively. The specificity of 3D OCT and digital radiography was 0.84 and 0.77, respectively, and no significant difference was observed between the values (p > 0.05). Diagnostic accuracy as the agreement with the CLSM histological view showed significantly higher results for 3D OCT than digital radiography ( Table 3, p < 0.05).
Sensors 2020, 20, x FOR PEER REVIEW 5 of 13 dental radiography, it was 0.56, 0.52, and 0.57, respectively. The specificity of 3D OCT and digital radiography was 0.84 and 0.77, respectively, and no significant difference was observed between the values (p > 0.05). Diagnostic accuracy as the agreement with the CLSM histological view showed significantly higher results for 3D OCT than digital radiography (Table 3, p < 0.05).  A tip of the cusp was slightly cracked (arrow). However, the anatomical shape of the cusp was maintained. The corresponding dynamic slicing 3D video is in Supplementary Materials: Video S1. The upper right is a cross-sectional image. Lower right is an en face image.        The cusp dentin was exposed to the tooth wear surface (arrow). Evidence of tooth demineralization or enamel cracks was not observed in this case. The corresponding dynamic slicing 3D video is in Supplementary Materials: Video S5. The image in the upper right is a cross sectional view. Lower right is an en face image.      Table 3. Diagnostic agreement with CLSM (weighted Kappa).

Discussion
Management of dental conditions is shifting from surgical treatment toward prevention and minimally invasive approaches [28]. Tooth wear is an irreversible process, and tooth loss is no exception for prevention and management. Recently, the prevalence of pathological tooth wear has increased in both the young generation and senior citizens, and patients are unaware of changes until the onset of symptoms such as DH or functional impairment. Accurate methodology for the monitoring of tooth loss is crucial for the prevention and management of tooth wear. In this study, 3D imaging of OCT was performed to evaluate the degree of tooth wear at the occlusal surface of extracted human molars. The null hypothesis of this study was rejected; 3D imaging of OCT for evaluation of tooth wear level improved the diagnosis compared with conventional methods using intraoral radiography in combination with visual inspection. The 3D images generated by OCT allowed for detailed examination of occlusal tooth wear by using multiplanar visualization of dynamic slicing (Figures 1-7, Videos S1-S7). The mathematical algorithm of 3D imaging installed in the OCT device appears to improve visualization of the remaining enamel and DEJ. A raw 3D image is generated as a 3D array of voxels or pixels with a grayscale range from 0 to 65,535 in a 16-bit pixel case. Three-dimensional imaging requires defined object boundaries, especially for the creation of 3D surface models by signal filtering. In the OCT system used in the current study, the noise is smoothed by using a Gaussian function on the image pixels, and the edge intensity and edge direction are extracted by an edge extracting filter. The image segmentation of pixel detection mathematically appears to minimize artifacts such as speckle noise, whereas images obtained by coherent imaging systems are inherently corrupted.
In this study, we classified the level of tooth wear into four categories; (1) slight enamel wear, (2) distinct enamel wear, (3) tooth wear with dentin exposure, and (4) tooth wear involving dentin. As tooth wear involving physiological conditions progresses naturally with aging, we presumed the slight enamel wear (score 1) as the physiological tooth wear level. Tooth wear having a score more than level 2 was presumed as at the pathological level and was evaluated by 3D OCT and digital dental radiography. Our results showed that 3D OCT could discriminate the distinct enamel wear and detect the involvement of dentin with significantly higher values of sensitivity, specificity, AUC, and agreement with histology than the radiography (Tables 1-3).
In OCT, amount of enamel loss could be estimated by the cross-sectional view of remaining enamel thickness, as the DEJ was clearly visualized as a landmark in all the OCT scans (Figures 1 and 2). Since intact enamel allows the OCT signal to penetrate deep with less scattering, the whole thickness of occlusal enamel could be monitored in this study, resulting in the higher agreement with histological findings (Table 3). These results correlate well with previous study findings measuring the remaining enamel thickness of tooth wear surfaces using cross-polarization (CP) OCT [25,26]. The CP-OCT measurement of remaining enamel thickness was reported to show excellent agreement with the µ-CT measurement [25,26]. Meanwhile, as the enamel thickness is highly location dependent, 3D OCT could visualize the remaining enamel thickness for the area of scanned tooth surface to find the position with the remaining enamel becoming thinner, without the need for additional scanning (Videos S1 and S2). Consequently, dentists allow finding of the location where therapeutic management is necessary, even for patients at the asymptomatic stage. In radiography, level of enamel wear is difficult to see because of the superimposition of dense buccal and lingual cusps, and visual inspection from the occlusal view estimating the topographical and color changes appears more informative in many cases.
Three-dimensional OCT could detect the presence of dentin exposure within the tooth wear surface with pinpoint accuracy (Figures 3-7, Videos S3-S7). Since dentin exposure may cause DH in response to mechanical and chemical stimuli during masticatory function, accurate diagnosis for dentin involvement appears beneficent for further maintenance and an intervention approach. Dentin contains 50 vol% of organic structure and scatters light, and signal attenuation of the OCT image through the dentin is higher than the enamel [29]. Optical dissimilarity between the two structures facilitates the discrimination of the tooth wear with and without dentin involvement. In this study, 3D OCT could clearly image the dentin exposure within the tooth wear surface, resulting in high diagnostic capacity and accuracy (Tables 1-3). OCT could also facilitate the characterization of exposed dentin surface changes, including the formation and thickness of transparent dentin [29].
In OCT, some of the samples with distinct tooth wear showed increased brightness at the superficial enamel or dentin, suggesting the presence of surface demineralization (Figures 4, 6 and 7) [16,17]. The incidence of demineralization was associated with tooth wear involving dentin (level 4) in many cases. Meanwhile, as the history of extracted teeth was unknown in this study, increasing the frequency of tooth demineralization with the advancement of tooth wear level may suggest the risk of erosion and ETW to facilitate tooth loss to the pathological level. Moreover, it is noteworthy that OCT could detect the enamel crack associated with tooth wear (Figure 6). This finding also showed increasing frequency with the advancement of tooth wear level. In this study, 11 cases out of 29 samples of level 4 dentin exposure showed distinct enamel crack penetrating into DEJ, and 2 cases of 29 samples showed slight enamel crack limited within the enamel thickness.
It is widely recognized that DEJ plays a key role as a damper to transfer loads from the enamel to dentin with different mechanical characteristics. Enamel cracks do not propagate into dentin due to the DEJ, which suppresses crack growth [30][31][32]. On the other hand, enamel tufts are hypocalcified wavy tissue near DEJ and are believed to be the source of cracks during heavy or continuous loading [33,34]. Several lines of evidence showed that dentin is more susceptible than enamel to erosion and abrasion alone or combined [35]. Although the critical pH at which oral fluid is saturated with calcium and phosphate from the tooth surface is not a constant, that for enamel is generally 5.5, yet the critical pH for dentin is 6.5 [36][37][38]. Therefore, once the dentin is exposed, the surface is vulnerable to erosion and can be eroded faster than the enamel. Moreover, the demineralization process can take place along the DEJ to cause an interfacial gap between the enamel and dentin ( Figure 6). Since tooth wear involving dentin reveals DEJ to the oral environment, the superficial enamel adjacent to the dentin exposure site may lose the underlying support of dentin due to the erosion along DEJ. The presence of superficial demineralization and enamel cracks at the tooth wear surface may accelerate the lesion progress and suggest the difficulty of clinical management and intervention for pathological tooth wear. These results also imply that subsurface demineralization and cracks should be taken into consideration for the prognosis of restorations for tooth wear surfaces. Consequently, accurate diagnosis of occlusal tooth wear regarding the depth and subsurface condition appears crucial for the determination of therapeutic approach. From this viewpoint, OCT appears to be a suitable method to monitor and manage the tooth wear.
Based on the results of this study, 3D OCT could provide images for diagnosing the level of tooth wear to improve the diagnostic accuracy of tooth wear compared with traditional methods using visual inspection and radiography. On the other hand, the rate of enamel loss in low pH was reported to speed up and occur faster than dentin, especially under the condition simulating mechanical action and gastric reflux [39,40]. Consequently, accurate diagnosis of early signs for tooth wear and ETW is essential to prevent further loss of tooth structure and pathological tooth wear. Within the limitation of this ex vivo study, 3D OCT can potentially help in providing valuable information for the remaining enamel thickness, amount of dentin exposure as well as enamel crack. Further study is necessary for clinical applications of OCT for tooth wear detection and monitoring, and for the management of pathological tooth wear.

Conclusions
3D OCT could image the degree of enamel loss and dentin exposure for occlusal tooth wear without the risk of X-ray exposure. The presence of tooth demineralization and enamel cracks involved in the tooth wear surfaces were also imaged in OCT. Three-dimensional OCT appears to be a safer option for the monitoring and management of tooth wear to prevent further tooth loss at the pathological level. Future clinical studies are required to image pathological tooth wear in vivo.