Comparing the diagnostic performance of a fluorescence-based caries detection tool of an intra-oral scanner to the gold-standard visual and radiographic examination

Abstract

Background: Intraoral scanners were originally developed for taking virtual impressions of teeth which are used to design and fabricate prosthodontic and orthodontic appliances. Recently, caries detection features have been included, but research regarding their diagnostic performance is still lacking, especially for the primary dentition. Objectives: The aims of this study were to determine the diagnostic performance of an intra-oral scanner with fluorescence technology (IOSFT) for caries detection on all the surfaces of premolars and molars in both the primary and permanent dentition compared to the gold-standard visual examination (for occlusal and smooth surfaces) and Bitewing examination (for interproximal surfaces). Fluorescence will be quantified using mean pixel values and then compared to visual examination for occlusal surfaces. The final aim was to investigate participant acceptance and compliance with IOSFT for caries detection. Materials and Methods: A total of 78 participants who consented were recruited from the University of Malta Dental teaching clinic according to a set of inclusion and exclusion criteria. 4283 permanent premolar and molar surfaces and 970 primary molar surfaces were cleaned with a prophy brush, water and floss and examined visually using ICDAS criteria. Bitewing (BW) radiographs were taken for each participant in a standardised method using the same film sizes, holders and x-ray machine settings. Participants were then scanned with the IOSFT. Following this, an anonymous questionnaire regarding the participants’ experience with the scanning procedure was distributed. The caries detection output on the software associated with the IOSFT was compared/correlated to the visual ICDAS scores obtained clinically for occlusal and smooth surfaces (buccal and lingual); whilst the IOSFT caries detection output was compared/correlated to radiographic ICDAS scores from BW radiographs for interproximal (mesial and distal) surfaces. The diagnostic performance of the IOSFT when compared to the gold-standard visual and radiographic caries detection was determined by calculating the sensitivity, specificity and accuracy values for each tooth surface. All these calculations were carried out for both the primary and the permanent dentitions and the IOSFT performance was compared between the two types of dentitions. Furthermore, fluorescence values on the occlusal surfaces of the primary and permanent teeth was quantified using mean pixel values. Standardised screen captures of the IOSFT software fluorescence output were exported to ImageJ (an image editing and analysis software) and the mean fluorescence pixel values (FPV) were measured for each occlusal surface in the study. This value was compared and correlated to the visual and radiographic ICDAS scores as well as to the IOSFT automated caries detection output. Results: The Shapiro-Wilk test was used to assess the data and normality was rejected (p<0.001). Non-parametric tests were used to investigate correlations and comparisons between variables. Out of the total number of surfaces, 487 permanent and 223 primary tooth surfaces had dental caries. There was a significant moderate correlation (r=0.4) between ICDAS/visual categories and IOSFT automated caries detection (ACD) categories for unrestored permanent occlusal surfaces (p<0.001). The correlation was stronger for primary occlusal surfaces (r=0.58, p<0.001). For both primary and permanent unrestored occlusal surfaces, IOSFT ACD accuracy was higher for irreversible lesions when compared to reversible ones (permanent occlusal 81% vs 37.7%; primary occlusal 88.7% vs 54.7%). In both cases, specificity was higher for irreversible lesions but sensitivity was higher for reversible ones. There was no correlation between IOSFT ACD categories and ICDAS/visual categories for restored occlusal surfaces (p>0.05). When unrestored smooth surfaces were investigated, weak significant correlations were found between ICDAS/visual categories and IOSFT ACD categories for both permanent (r=0.19) and primary (r=0.29) tooth surfaces (p<0.001), with results slightly favouring primary teeth. Good results were achieved for accuracy of IOSFT ACD in both dentitions. Accuracy results for permanent smooth surfaces were 93.5% for irreversible lesions and 77.1% for reversible lesions whilst the respective results for primary smooth surfaces were 95.1% and 79.9%. For interproximal lesions, correlations between BW ICDAS scores and IOSFT ACD scores were weak for permanent surfaces (r=0.07, p<0.019) and moderate for primary surfaces (0.413, p<0.001). Diagnostic accuracy measures for IOSFT ACD was also marginally higher for both irreversible and reversible lesions in primary (84.3% vs 49.5%) when compared to permanent (83.5% vs 40.42%) surfaces. Sensitivity was higher for reversible lesions whilst specificity was higher for irreversible lesions in both dentitions. Mean fluorescence pixel values (FPV) were used to quantify fluorescence on permanent and primary occlusal surfaces. A Spearman correlation between ICDAS/visual categories and mean FPV revealed a significant weak inverse correlation for unrestored permanent teeth (r= -0.13, p<0.001), and a significant strong inverse correlation for unrestored primary teeth (r= -0.52, p<0.0001). Kruskal-Wallis comparisons between mean FVP and three categories of visual examination also revealed significant differences for both the permanent (p=0.004) and primary (p<0.001) unrestored occlusal surfaces. Statistically significant differences in mean FPV were found between sound primary and permanent occlusal surfaces (p<0.0001). Significant correlations were also found between mean FPV and participant age. There was a weak significant inverse correlation for the variables in permanent teeth (r= -0.27, p<0.0001) and a moderate significant correlation in primary teeth (r=0.31, p<0.0001). No significant correlation was found between mean FPV and ICDAS/visual categories for restored occlusal surfaces (secondary caries) (p<0.050), however, a significant difference was found in the mean FPV between different restorative materials (Kruskal Wallis test, p<0.0001). Scanning time and number of images decreased significantly with operator training time (p<0.001), but both time taken (p<0.001) and number of images (p<0.0001) were significantly higher for the maxilla (compared to the mandible) and for older participants (p<0.0001). The questionnaire revealed that the vast majority of participants (87%) found the scanning procedure comfortable and acceptable but there were statistically significant differences in discomfort experienced in different areas of the mouth, with the most uncomfortable area scanned being the distal surfaces of the upper molars (p<0.001). Conclusions: The correlations as well as the diagnostic performance of the IOSFT compared to the gold-standard clinical caries detection methods (visual for occlusal and smooth surfaces and Bitewing radiographs for interproximal surfaces) was significantly beter for all primary surfaces, especially so for occlusal surfaces and irreversible lesions. Similar results were found for mean (FPV) of occlusal surface caries when compared and correlated with visual examination categories. IOSFT performance for secondary caries detection was unfavourable. Participant acceptance when scanning for caries detection was very high and considered the procedure to be comfortable in most areas in the mouth.