Deep Learning for the Radiographic Detection of Apical Lesions

 Published online: May 31, 2019

Abstract

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.
Continue reading

Does selective carious tissue removal of soft dentin increase the restorative failure risk in primary teeth?: Systematic review and meta-analysis.

J Am Dent Assoc. 2019 May 27. pii: S0002-8177(19)30134-5. doi: 10.1016/j.adaj.2019.02.018. [Epub ahead of print]

xAbstract

BACKGROUND:

The authors conducted a systematic review of randomized controlled trials comparing the risk of experiencing restoration failure in primary teeth after complete and selective carious tissue removal of soft dentin.

METHODS:

The authors searched electronic databases (PubMed [MEDLINE], Scopus, Cochrane Central Register of Controlled Trials) and the ClinicalTrials.gov Web site with manual searching and cross-referencing for trials reporting restoration failure after follow-up of 6 months or longer. Two reviewers independently selected studies, extracted data, and assessed the risk of bias and quality of evidence using the Grading of Recommendations Assessment, Development, and Evaluation approach. The authors performed intention-to-treat and per-protocol meta-analyses and calculated odds ratios (OR) as effect estimates in the random-effects model.

RESULTS:

From 327 potentially eligible studies, the authors selected 23 for full-text screening and included 4. Results showed increased risk of experiencing restoration failure (intention-to-treat analysis, OR [95% confidence interval] 1.74 [1.01 to 3.00], and per-protocol analysis, OR [95% confidence interval] 1.79 [1.04 to 3.09]) after selective carious tissue removal of soft dentin. The risk of bias was high, and the quality of evidence was low.

CONCLUSIONS:

Selective carious tissue removal of soft dentin may increase the risk of experiencing restoration failure in primary teeth. However, the evidence level is insufficient for definitive conclusions.

PRACTICAL IMPLICATIONS:

Patients with restorations performed after selective carious tissue removal of soft dentin should have shorter recall visit intervals to evaluate the restorations’ quality and control caries disease, allowing for more conservative approaches, such as repair, in cases of defective restorations.

Continue reading

Association between periodontitis and cognitive impairment: analysis of national health and nutrition examination survey (NHANES) -III.

J Clin Periodontol. 2019 Jun 1. doi: 10.1111/jcpe.13155. [Epub ahead of print]

Abstract

OBJECTIVES:

Periodontitis has been hypothesized as being one of the most common potential risk factors for the development of dementia and cognitive impairment. In order to investigate the relationship between periodontitis and cognition impairment, the National Health and Nutrition Examination Survey (NHANES) database was analyzed after adjusting for potential confounding factors, including age and other systemic co-morbidities.

MATERIALS AND METHODS:

In total, 4663 participants aged 20 to 59 years who had received full mouth periodontal examination and undergone the cognitive functional test were enrolled. The grade of periodontal disease was categorized into severe, moderate, and mild. Cognitive function examinations, including the Simple Reaction Time test (SRTT), Symbol Digit Substitution Test (SDST), and Serial Digit Learning Test (SDLT), were adopted for the evaluation of cognitive impairment.

RESULTS:

The subjects with mild and moderate to severe periodontitis had higher SDLT and SDST scores, which indicated decreased cognitive function, compared to the healthy group. After adjusting for demographic factors, education, smoking, cardiovascular diseases, and laboratory data, periodontitis was significantly correlated with elevated SDST and SDLT scores (P values for trend = 0.014 and 0.038, respectively) by generalized linear regression models.

CONCLUSION:

Our study highlighted that periodontal status was associated with cognitive impairment in a nationally representative sample of US adults.

Continue reading

Smartphones and tooth brushing: content analysis of the current available mobile health apps for motivation and training

Abstract

Introduction

Smartphone mobile app is an innovative concept for health behaviour‐based interventions.

Aim

The present study aimed to analyse apps developed for smartphones that promote tooth brushing amongst children using the Coventry, Aberdeen, and London-Refined (CALO-RE) taxonomy for behaviour change.

Materials and methods

Tooth brushing apps available in English and free to download that purported to assist with brushing were searched on the Apple app store using search terms based on Boolean logic and included AND combinations for keywords tooth brushing, children, toothbrush and motivation in the health and fitness category; six apps met the inclusion criteria and were downloaded. The behaviour change taxonomies were assessed individually for each app and scored as per coding and analysed for presence or absence.

Results

Only three of the behaviour change taxonomies were present in all apps, i.e. information provision (general), goal setting (behaviour) and prompt practice. Setting graded tasks, self-monitoring of behavioural outcome, demonstration of behaviour, prompt use of imagery and time management were included in four out of six apps.

Conclusion

The present study explores a new arena for oral healthcare motivation and prevention in children through the use of mobile phone apps.
Continue reading

Changes in oxygen and carbon dioxide in the genesis of sleep bruxism: a mechanism study

Journal of Prosthodontic Research

Available online 31 May 2019

Abstract

Purpose

Rhythmic masticatory muscle activity (RMMA), a biomarker of sleep bruxism (SB), has been associated with mild hypoxia and/or big breaths in some adults with non-sleep-disordered breathing. The purpose of this study was to investigate that concurrent oxygen and carbon dioxide fluctuations are among the physiological variables that contribute to RMMA onset.

Methods

Twelve subjects (5 female, 7 male, mean age: 43 ± 11) underwent polysomnography recording in a sleep laboratory. RMMA index and apnea-hypopnea index were calculated. Oxygen saturation (SpO2) was estimated by finger pulse oximeter and end-tidal CO2 (ETCO2) by nasal airflow cannula before and after RMMA onset. Given the expected response time delay between actual arterial hypoxemia and fingertip pulse detection, we adjusted the SpO2 desaturation onset to the onset of masseter muscle activity using a 17 s criterion based on ETCO2 shifts.

Results

SpO2 was slightly but significantly lower than at baseline (max: −0.6%) in the 6–4 s before RMMA onset and significantly higher in the 6–18 s after onset (0.9%; p < 0.05). Although ETCO2 before RMMA onset did not differ from baseline, it decreased at 8–10 s after onset (−1.7 mmHg: p < 0.05). No changes in SpO2 or ETCO2 in relation to RMMA onset reached a critical clinical threshold.

Conclusions

The mild transient hypoxia observed before RMMA onset was not associated with a change in ETCO2. The mild and brief oxygen fluctuations before RMMA onset may reflect a physiological response that seems to have little influence on SB genesis.
Continue reading

Electronic and paper mode of data capture when assessing patient‐reported outcomes in the National Dental Practice‐Based Research Network

First published: 02 June 2019

 

Abstract

Aim

Our objectives were to describe the approach used in the National Dental Practice‐Based Research Network to capture patient‐reported outcomes and to compare electronic and paper modes of data capture in a specific network study.

Methods

This was a prospective, multicenter cohort study of 1862 patients with dentin hypersensitivity. Patient‐reported outcomes were assessed based on patients’ perception of pain using Visual Analog Scales and Labeled Magnitude scales at baseline and at 1, 4 and 8 weeks post‐baseline.

Results

Eighty‐five percent of study patients chose to complete follow‐up assessments via an electronic mode; 15% completed them via a paper mode. There was not a significant difference in the proportions of patients who completed the 8‐week assessment when comparing the electronic mode to the paper mode (92% vs. 90.8%, P = 0.31, Rao‐Scott clustered χ2‐test).

Conclusion

The electronic mode of data capture was as operational as the traditional paper mode, while also providing the advantage of eliminating data entry errors, not involving site research coordinators in measuring the patient‐reported outcomes, and not incurring cost and potential delays due to mailing study forms. Electronic data capture of patient reported outcomes could be successfully implemented in the community dental practice setting.

Continue reading