Clustering effects of oral conditions based on clinical and radiographic examinations.

OBJECTIVES: The intra-class correlation coefficient (ICC) is a measure of intra-subject clustering effects. A priori estimates of the ICC and the associated design effect (DE) are required for sample size estimation in clustered studies, and should be considered during their analysis, too. We aimed to determine the clustering effects of carious lesions, apical lesions, periodontal bone loss, and periodontal pocketing, assessed in clinical or radiographic examinations. METHODS: A subsample of patients (n = 175) enrolled in the fifth German Oral Health Study provided data on clinically determined carious teeth (i.e., with untreated carious lesions, WHO method) as well as teeth with periodontal pocketing (i.e., with maximum probing-pocket-depths ≥ 4 mm). A sample of panoramic radiographs (n = 85) from randomly chosen patients, examined from 2010 to 2017 at the Charité dental hospital, provided data on radiographically determined carious teeth (i.e., with lesions extending into dentine or enamel), teeth with apical lesions (determined by dentists via majority vote), and teeth with periodontal bone loss (≥ 20% of root-length). The ICC and its 95% confidence interval (95% CI) were determined. RESULTS: There were 3839 and 1961 teeth assessed in clinical and radiographic evaluations, respectively. For clinically or radiographically determined carious lesions, the ICC (95% CI) was 0.20 (0.16-0.24) or 0.19 (0.14-0.25), respectively. For clinical pocketing or radiographic bone loss, the ICC was 0.40 (0.35-0.46) or 0.30 (0.24-0.38), respectively. The lowest ICC was found for apical lesions at 0.08 (0.06-0.13). CONCLUSIONS: The ICC varied between assessment methods and conditions. Clustered trials should account for this during study planning and data analysis. CLINICAL RELEVANCE: Within the limitations of this study, and considering the risk of selection bias and the limited sample sizes of both datasets, clustering effects were substantial but varied between dental conditions. Studies not accounting for this during planning and analysis may yield misleading estimates if clustering is present.

Deep Learning for the Radiographic Detection of Periodontal Bone Loss.

We applied deep convolutional neural networks (CNNs) to detect periodontal bone loss (PBL) on panoramic dental radiographs. We synthesized a set of 2001 image segments from panoramic radiographs. Our reference test was the measured % of PBL. A deep feed-forward CNN was trained and validated via 10-times repeated group shuffling. Model architectures and hyperparameters were tuned using grid search. The final model was a seven-layer deep neural network, parameterized by a total number of 4,299,651 weights. For comparison, six dentists assessed the image segments for PBL. Averaged over 10 validation folds the mean (SD) classification accuracy of the CNN was 0.81 (0.02). Mean (SD) sensitivity and specificity were 0.81 (0.04), 0.81 (0.05), respectively. The mean (SD) accuracy of the dentists was 0.76 (0.06), but the CNN was not statistically significant superior compared to the examiners (p = 0.067/t-test). Mean sensitivity and specificity of the dentists was 0.92 (0.02) and 0.63 (0.14), respectively. A CNN trained on a limited amount of radiographic image segments showed at least similar discrimination ability as dentists for assessing PBL on panoramic radiographs. Dentists' diagnostic efforts when using radiographs may be reduced by applying machine-learning based technologies.

Deep Learning for the Radiographic Detection of Apical Lesions.

INTRODUCTION: We applied deep convolutional neural networks (CNNs) to detect apical lesions (ALs) on panoramic dental radiographs. METHODS: Based on a synthesized data set of 2001 tooth segments from panoramic radiographs, a custom-made 7-layer deep neural network, parameterized by a total number of 4,299,651 weights, was trained and validated via 10 times repeated group shuffling. Hyperparameters were tuned using a grid search. Our reference test was the majority vote of 6 independent examiners who detected ALs on an ordinal scale (0, no AL; 1, widened periodontal ligament, uncertain AL; 2, clearly detectable lesion, certain AL). Metrics were the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and positive/negative predictive values. Subgroup analysis for tooth types was performed, and different margins of agreement of the reference test were applied (base case: 2; sensitivity analysis: 6). RESULTS: The mean (standard deviation) tooth level prevalence of both uncertain and certain ALs was 0.16 (0.03) in the base case. The AUC of the CNN was 0.85 (0.04). Sensitivity and specificity were 0.65 (0.12) and 0.87 (0.04,) respectively. The resulting positive predictive value was 0.49 (0.10), and the negative predictive value was 0.93 (0.03). In molars, sensitivity was significantly higher than in other tooth types, whereas specificity was lower. When only certain ALs were assessed, the AUC was 0.89 (0.04). Increasing the margin of agreement to 6 significantly increased the AUC to 0.95 (0.02), mainly because the sensitivity increased to 0.74 (0.19). CONCLUSIONS: A moderately deep CNN trained on a limited amount of image data showed satisfying discriminatory ability to detect ALs on panoramic radiographs.