Diagnosis of Pit-and-fissure Caries Using Three-dimensional Scanned Images
Diagnosis of the extent of pit-and-fissure caries has been subjective and thus difficult to teach and categorize for treatment planning. This in vitro study compares occlusal caries diagnosis of extracted posterior teeth (n=49) using three-dimensional (3D) scanned images vs visual examination, according to the International Caries Detection and Assessment System (ICDAS). The surfaces chosen for study represent all ICDAS classifications. Five experienced restorative faculty members examined scanned images for 60 seconds from a standardized series of views of each surface and scored them independently. One month later, the same teeth were examined visually by the same five raters with magnification and LED headlamps, with compressed air available. Intrarater and interrater agreement and validity were assessed using intraclass correlation coefficients (ICCs). The ICCs, ranging from 0.90 to 0.93, indicated excellent agreement between and within raters and between the raters and the gold standard ICDAS determination. This suggests that both photographs and 3D scans of pits and fissures are equally effective in diagnosing caries.SUMMARY
INTRODUCTION
Clinicians have long noted the difficulty of accurately assessing the extent of pit-and-fissure caries.1 Carious lesions exist on a continuum of severity from early decalcification to complete destruction of the clinical crown and root. A standard descriptive framework that reasonably predicts the amount of carious tissue destruction is accordingly required to develop an appropriate treatment plan. While the amount of physical cavitation of the lesion has been proposed2 as the gold standard in evaluating smooth surface caries, and although the extent of damage can be evaluated on standard radiographs, predicting caries level in occlusal pit-and-fissure caries has proven more problematic. It has been suggested that extensive fluoridation has created enamel that is more resistant to acid attack, delaying frank cavitation even when dentin caries has progressed.3 The traditional technique of caries detection taught to generations of clinicians involves forcing a sharp explorer into the suspect fissure and testing for resistance to withdrawing (“a stick”). This technique has been long discredited, but it still persists in practice.4,5 The standard operative classification proposed by GV Black in the 1800s is not based upon describing the lesion extent or activity. Instead, it references the planned location of the restoration.
An alternate system using visual diagnostic criteria for caries is defined in the International Caries Detection and Assessment System (ICDAS II), which was developed as a tool for epidemiology and research and has also been accepted as a formal curricular framework for teaching in the United States.6 This system has been validated through comparison to histologic, radiologic, and fluorescence-based findings and was shown to correlate more highly than radiographs in treatment decisions made by experienced practitioners.7-11,12 The basis of the ICDAS system is careful visual evaluation of the fissure system with magnification to 2.5×, as stronger magnification has been shown to decrease the specificity of examination to unacceptable levels.13 Intraoral photographs of teeth have been compared with visual appearance of teeth on five- to 11-year-olds and have been found to be equivalent for teaching, epidemiologic, and research purposes.14 Visual exam with ICDAS has also been shown to be superior to radiographs in detecting occlusal caries.11
The widespread use of three-dimensional (3D) digital imaging devices in clinical care for documentation, patient education, and creation of impressions has created an opportunity to use these images for diagnosis as well. One pilot study12 investigating the feasibility of teledentistry in remote pediatric dental evaluation found no difference in dmft/DMFT scores between clinical exams and photos made with a two-dimensional intraoral camera. Electronic 3D images have potential for use in creating realistic teaching cases for students, in which the images could be manipulated in visual space. Since images are routinely transmitted electronically for laboratory fabrication of restoration it should be possible to use them to document caries for insurance purposes, especially in the early stages, during which they are more accurate than radiographs.12 To date, no study has looked at 3D imaging when compared to clinical appearance in ICDAS scoring. That was the purpose of this study.
METHODS AND MATERIALS
Sample Selection and Classification
A sample size calculation indicated that, with 95% confidence, having five raters examine 49 teeth would yield estimated intrarater (IAR) and interrater (IER) agreement values that were within 0.10 of the true values, assuming that the true repeatability was 0.71 or greater. Thus, in this institutional review board–exempt study, seven permanent teeth from each ICDAS category (see Table 1; 49 teeth total) were selected from a pool of recently extracted posterior teeth without restorations or fluorosis.
The teeth were cleaned of calculus and debris using a scaler and toothbrush under running water and were then stored in 0.9% sodium chloride/0.2% sodium azide solution at 36°F. Using 2.5× magnifying dental loupes (Designs for Vision Inc, Long Island, NY, USA) and headlamps (Ultra-Light Optics, Fountain Valley, CA, USA), the teeth were then categorized by two senior educators experienced in teaching ICDAS. The consensus opinion of the ICDAS classifications was considered the “gold standard” reference classification for the study.
3D Scanning of Teeth
Using the 3D color imaging mode of a high-definition hand-held intraoral scanner (3Shape, Copenhagen, Denmark) images were captured per the ICDAS protocol of first examining the surface wet, then drying it with a five-second air stream, and then noting the changes between wet and dry. Six images were captured of each tooth (three wet and three dry), one of the occlusal (Figure 1), one from a slight lingual tilt (Figure 2), and one from a slight facial tilt to simulate a clinical situation. Images were processed with 3Shape Trios Version 1.3.4.5 and stored as screen shots on a MacBook Pro Retina (Apple Inc, Cupertino, CA, USA).
Citation: Operative Dentistry 43, 3; 10.2341/17-076-L
Citation: Operative Dentistry 43, 3; 10.2341/17-076-L
Data Collection
Five clinician educators experienced with the ICDAS method successfully completed online ICDAS training (www.icdas.org) for calibration. They were brought together in a room with normal lighting and shown images on a 27-inch iMac 5K (Apple) to simulate a clinical evaluation. Each tooth was represented by 3D images captured wet and dry. To simulate the clinical evaluation, the wet image was viewed first for 20 seconds by all evaluators, with a standardized rotation of the 3D image of viewing from the occlusal straight on for 10 seconds, slightly to the buccal for five seconds, and slightly to the lingual for five seconds. The dry surface scan was then presented for 40 seconds with the same standardized rotation for 10 seconds for each of the three angles and a final 10 seconds on the straight occlusal image. Each evaluator independently scored the surface. This was designated the “digital” approach.
The teeth were then stored in separate numbered vials in sodium azide solution at 36°F. They were randomized according to a schedule maintained by an investigator who did not participate in the examination process.
Approximately one month later, the same investigators independently evaluated all of the teeth visually for one minute per tooth. Teeth were presented in separate cups in water, so they were examined first wet, then after a five-second air stream using a head lamp and 2.5× magnification. Each was assigned an ICDAS score based on this examination. This was designated the “analog” approach.
Statistical Analysis
To assess reliability, both IAR agreement and IER agreement were examined. IAR measures the extent to which each rater tends to agree with himself between the analog and digital approaches, regardless of individual tooth or ICDAS classification. Intraclass correlation (ICC) was used to assess the IAR for all raters combined. IER measures the extent to which the five raters tend to agree with each other, regardless of method (analog vs digital), individual tooth, or ICDAS classification. The ICC was used to assess the IER among the five raters for the analog and digital methods combined and separately for the analog and digital methods, regardless of individual tooth or ICDAS classification.
To assess validity, the ICC was used to measure agreement between the analog result and the gold-standard ICDAS classification and between the digital result and the gold-standard ICDAS classification, regardless of rater or individual tooth. Reliability and validity were assessed by calculating the appropriate ICCs after combining the classification results obtained by the five raters with the gold standard results for all 49 teeth, which consisted of seven teeth in each of the ICDAS classes 0-6.
Evaluation of the IER and IAR within each ICDAS classification was not an objective of the study and could not be attempted because the number of specimens (n=7) was too low, relative to the number of raters, to yield a meaningful assessment of reliability.
With regard to interpretation, the ICC as an agreement coefficient ranges between a maximum value of 1, which indicates “perfect agreement,” and a minimum value of 0, indicating “no agreement.” A value of the ICC between 0.75 and 1.00 indicates “excellent” agreement, values between 0.40 and 0.74 indicate “fair to good” agreement, and values less than 0.40 indicate “poor” agreement.15 The method of Gilder and others was used to estimate the ICC and to find approximate 95% confidence intervals (CIs) for all agreement coefficients.16 The CI provides information on how precise the estimated agreement coefficient is: the narrower the interval, the more precise the agreement; the wider the interval, the less precise the agreement. All calculations were performed using SAS 9.4.
RESULTS
Table 2 contains the IER and IAR agreement coefficients based upon the full ICDAS classification system. For this analysis, there were five raters, 49 specimens (ie, teeth), and two “trials” (analog and digital). All of the ICC values in Table 1 indicate “excellent” IER and IAR agreement and, hence, excellent reliability. The CIs are quite narrow, indicating a high degree of precision in the estimation of the agreement coefficients.
Table 3 contains the ICC values for agreement between results obtained using the analog method and the teeth classified using the gold-standard ICDAS determination and agreement between results obtained using the “digital” method and the gold-standard ICDAS. All of the ICC values indicate excellent agreement for both the analog and digital methods, and the CIs are quite narrow, thus establishing excellent validity for both methods.
DISCUSSION
Using visual examination of pit-and-fissure caries has been shown6,8,18 to be clinically appropriate and effective. To describe and categorize visual evidence, two classification systems are currently in use. In addition to ICDAS, the American Dental Association has published a simplified version of the ICDAS, known as the Caries Classification System (CCS), which combines similar ICDAS criteria to create four CCS categories instead of the seven used in the ICDAS. Diagnosis with visual criteria is currently included in the core curriculum framework in cariology for US dental schools, with two systems, ICDAS and CCS, currently proposed.19,20 For the purposes of this study, ICDAS was chosen because the investigators considered it more detailed than CCS.
This study indicates that caries detection using ICDAS visual criteria with 3D scans can be as accurate as direct visual assessment. Valid and reliable diagnosis of occlusal carious lesions is an important first step in appropriate treatment planning. ICDAS visual criteria can provide information on the level of disease present clinically that, when taken in the context of the caries risk assessment of diet, salivary flow, and recent history of carious lesions, provides guidance for treatment of the patient.11,12 If 3D images are validated as accurate representations of disease state then digital treatment planning may become a reality, with digital radiographs, occlusal scans, and digital models being used to document extent of disease.
When the entire data set of 49 teeth was analyzed, there was excellent IER and IAR agreement and excellent agreement of the ICDAS gold standard with both the analog and digital methods. This study showed that experienced evaluators are as accurate in classifying pit-and-fissure caries with scanned images as they are with direct visual exam, and this study also showed more overall consistency among raters than has been associated with most previous studies. The authors detected one outlier in each of the seven ICDAS categories, defined as the median score for that tooth differing by at least one ICDAS classification from the gold standard result. Because of this discrepancy, the authors considered eliminating these teeth from the analysis, but the agreement results after removing them (results not shown) did not differ in any meaningful way from the results based on all 49 teeth.
In the educational setting, scans may convey more information to students in teaching how to identify caries using ICDAS than other modalities, since several scan views can be evaluated to teach in a more realistic manner. There may also be a role for scans in verifying clinical data for insurance purposes, since scanned images can be transmitted digitally via e-mail. Further in vivo clinical studies would be helpful to confirm these findings.
CONCLUSION
Within the limitations of this in vitro study, caries classification of pits and fissures can be evaluated as accurately with 3D scanned images as with direct visual examination in vitro.
Occlusal view, dry. Scanned three-dimensional (3D) image of the occlusal of an ICDAS 5 lesion. Note the cavitation as well as areas of surface decalcification and shadowing in other areas visible in the scan. Original scan is in color.
Lingual tilt view, dry. Same tooth viewed from the lingual to show additional features of cavitated area and surface anatomy.
Contributor Notes
Martha G Brackett, DDS, MSD, Dental College of Georgia at Augusta University, Restorative Sciences, Augusta, GA, USA