Comparing potential early caries assessment methods for teledentistry

Abstract

Background: Optical caries detection has the potential to be incorporated in telehealth medicine for preventive dental screening. The objective of this study was to evaluate and compare visible and near infrared detection methods for identifying early non-cavitated ex vivo occlusal demineralization.

Methods: Six blinded examiners were used to compare the accuracy of the following three examinations in detecting occlusal demineralization: Midwest Caries ID (MID), visual photographic examination (CAM) and Cross Polarization Optical Coherence Tomography (CP-OCT). For each diagnostic method, two examiners assessed the extracted tooth samples 1-2 weeks apart. Teeth were then sectioned and lesion depth was confirmed (n = 42) by a blinded histological examination using a glycol based caries indicator dye. The sensitivity (Sen), specificity (Sp), Intraclass Correlation Coefficient (ICC), and Area under the Receiver Operator Curve (AUC) were calculated.

Results: For detecting any demineralization versus sound pit and fissure enamel, the mean Sen/Sp found was 46.9/85.0 for MID, 80.5/52.5 for CAM, and 83.4/45.0 for CP-OCT. For detecting non-cavitated demineralization that progressed into the dentin, the mean Sen/Sp found was 17.3/88.0 for MID, 48.0/57.8 for CAM, and 44.2/72.7 for CP-OCT. AUC values were statistically significant (P < 0.05) in three out of four examiner assessments when MID and CP-OCT were used to detect any demineralization. AUC values were significant for a single CAM examination. When assessing deeper non-cavitated lesions, none of the assessment methods were able to yield AUC values that were significantly different than a random 'coin flip' test. When examining reliability, MID demonstrated the highest ICC score (0.83) and CP-OCT had the lowest (0.49).

Conclusion: Although MID and CP-OCT were useful in detecting the presence of demineralization, examiners were not able to utilize these devices to adequately assess the depth of the demineralization. This study found that MID and CP-OCT did not have markedly superior diagnostic values from simple CAM assessment for use in teledentistry.

Figure 1

Each tooth had two composite reference points (r1 and r2). The test area was the fissure area between the points. A) All CAM images were assessed Score ‘2’. All MID Scorings were ‘0’. B) CP-OCT images were taken before tooth sectioning. CP-OCT images are presented in grayscale. Black areas have low scattering and white areas have high scattering. All CP-OCT scorings were ‘2’ based on the fissure area near r1 (arrow). C) Histological evaluation using caries indicator dye confirmed demineralization extending into enamel and dentin with Score of ‘2’ (arrow).

Figure 2

Ex vivo experimental design. A total of seven blinded examiners (E) were used for the study. Examiner 1 (E1) initially sorted the tooth samples, collected one of the MID readings and obtained CAM and CP-OCT images, and compiled the results. E1 was blinded to the other examiners (E2-E7) assessments prior to recording the MID readings. E2-E7 were blinded. E5 and E6 did not have access to any visual images of the tooth samples and made assessments solely on the CP-OCT images provided.