Objectives This ex vivo study aimed to compare the accuracy in detection of interproximal natural carious lesions between intraoral (iBWR) and extraoral bitewing radiographs (eBWR) using a multi-observer design and a rigorous gold standard. Materials and methods Eighty extracted teeth (40 premolars, 40 molars) were arranged in anatomical sequence within a simulated jaw composed of PMMA and modified gypsum, with an emphasis on creating natural interproximal contacts. Approximately 50% of the teeth exhibited enamel caries, while the remaining 50% were caries-free. Image acquisition was performed using a custom-designed PMMA phantom. iBWR were obtained with a CMOS intraoral sensor (XIOS XG Supreme, Sirona Dental Systems, Bensheim, Germany), and eBWR with a digital panoramic device (Orthophos SL 3D, Dentsply Sirona, Bensheim, Germany). Twenty-seven licensed dentists assessed caries presence and depth on 120 approximal surfaces (each surface assessed twice using both modalities) using a 5-point confidence scale and a 4-point lesion depth scale. Observers were blinded to the true caries status, which was determined through histological serial sectioning and brightfield microscopy. Diagnostic accuracy was evaluated via ROC analysis, with Youden’s index used to calculate sensitivity, specificity, predictive values, and likelihood ratios. Statistical analyses were conducted at a significance level of α = 0.05. Results Overall accuracy was higher for iBWR (Azpooled = 0.58) than for eBWR (Azpooled = 0.54). Both intra-rater (test-retest, eBWR <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="784_2025_6511_Article_IEq1.gif" Format="GIF" Height="12" Rendition="HTML" Resolution="72" Type="Linedraw" Width="13" /> </InlineMediaObject> <EquationSource Format="TEX">\(\:\rho\:\)</EquationSource> </InlineEquation>spearman = 0.44, iBWR <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="784_2025_6511_Article_IEq1.gif" Format="GIF" Height="12" Rendition="HTML" Resolution="72" Type="Linedraw" Width="13" /> </InlineMediaObject> <EquationSource Format="TEX">\(\:\rho\:\)</EquationSource> </InlineEquation>spearman = 0.48) as well as inter-rater reliability (mean ICC eBWR = 0.19, iBWR = 0.27) were low. For enamel caries detection, iBWR outperformed eBWR in terms of specificity and positive predictive values, while eBWR in the first reading round achieved significantly higher sensitivity. Conclusions Overall, our multi-observer ex vivo study using microscopy as ground truth revealed higher diagnostic accuracy for intraoral bitewing radiography as compared to its extraoral counterpart. Clinical Relevance Our results from a highly standardized study using a rigorous gold standard support the assumption that intraoral bitewing radiography still represents the radiographic state-of-the-art in interproximal caries detection. For minute enamel, diagnostic accuracy of both methods is just above random guessing.

Intraoral vs. extraoral bitewing radiography for approximal caries detection: A multi-observer ex vivo ROC study using thin-section microscopy as gold standard

期刊论文
出版模式：

开放获取
发表日期：

2025年9月26日

查看全文

Julia Caroline Quintus,
Ralf Kurt Willy Schulze

摘要

Objectives

This ex vivo study aimed to compare the accuracy in detection of interproximal natural carious lesions between intraoral (iBWR) and extraoral bitewing radiographs (eBWR) using a multi-observer design and a rigorous gold standard.

Materials and methods

Eighty extracted teeth (40 premolars, 40 molars) were arranged in anatomical sequence within a simulated jaw composed of PMMA and modified gypsum, with an emphasis on creating natural interproximal contacts. Approximately 50% of the teeth exhibited enamel caries, while the remaining 50% were caries-free. Image acquisition was performed using a custom-designed PMMA phantom. iBWR were obtained with a CMOS intraoral sensor (XIOS XG Supreme, Sirona Dental Systems, Bensheim, Germany), and eBWR with a digital panoramic device (Orthophos SL 3D, Dentsply Sirona, Bensheim, Germany).

Twenty-seven licensed dentists assessed caries presence and depth on 120 approximal surfaces (each surface assessed twice using both modalities) using a 5-point confidence scale and a 4-point lesion depth scale. Observers were blinded to the true caries status, which was determined through histological serial sectioning and brightfield microscopy. Diagnostic accuracy was evaluated via ROC analysis, with Youden’s index used to calculate sensitivity, specificity, predictive values, and likelihood ratios. Statistical analyses were conducted at a significance level of α = 0.05.

Results

Overall accuracy was higher for iBWR (Az_pooled = 0.58) than for eBWR (Az_pooled = 0.54). Both intra-rater (test-retest, eBWR \(\:\rho\:\) _spearman = 0.44, iBWR \(\:\rho\:\) _spearman = 0.48) as well as inter-rater reliability (mean ICC eBWR = 0.19, iBWR = 0.27) were low. For enamel caries detection, iBWR outperformed eBWR in terms of specificity and positive predictive values, while eBWR in the first reading round achieved significantly higher sensitivity.

Conclusions

Overall, our multi-observer ex vivo study using microscopy as ground truth revealed higher diagnostic accuracy for intraoral bitewing radiography as compared to its extraoral counterpart.

Clinical Relevance

Our results from a highly standardized study using a rigorous gold standard support the assumption that intraoral bitewing radiography still represents the radiographic state-of-the-art in interproximal caries detection. For minute enamel, diagnostic accuracy of both methods is just above random guessing.