Teledentistry as a novel pathway to improve dental health in school children: a research protocol for a randomised controlled trial.

BACKGROUND: Despite great improvement in child oral health, some children subgroups still suffer from higher levels of dental caries. Geographic and socioeconomic barriers and the lack of access to dental care services are among common reasons for poor oral health in children. Historically in Australia, oral health therapists or dental therapists have been responsible for providing dental care for school children through the School Dental Services (SDS). The current SDS has been unable to provide sustainable dental care to all school children due to a reduction in workforce participation and limited resources. We propose a paradigm shift in the current service through the introduction of user-friendly technology to provide a foundation for sustainable dental care for school children. METHODS/DESIGN: We describe an ongoing parallel, two-armed, non-inferiority randomised controlled trial that compares routine and teledental pathway of dental care in children aged 4-15 years (n = 250). Participating schools in Western Australia will be randomly assigned to the control or teledental group, approximately three schools in each group with a maximum of 45 children in each school. All participants will first receive a standard dental examination to identify those who require urgent referrals and then their teeth will be photographed using a smartphone camera. At the baseline, children in the control group will receive screening results and advice on the pathway of dental care based on the visual dental screening while children in the teledental group will receive screening results based on the assessment of dental images. At 9 months follow-up, all participants will undergo a final visual dental screening. The primary outcomes include decay experience and proportion of children become caries active. The secondary outcomes include the diagnostic performance of photographic dental assessment and costs comparison of two pathways of dental care. DISCUSSION: The current project seeks to take advantage of mobile technology to acquire dental images from a child's mouth at school settings and forwarding images electronically to an offsite dental practitioner to assess and prepare dental recommendations remotely. Such an approach will help to prioritise high-risk children and provide them with a quick treatment pathway and avoid unnecessary referrals or travel. TRIAL REGISTRATION: Australian New Zealand Clinical Trials Registry, ACTRN12619001233112. Registered 06 September 2019.

A Deep Learning-Based System for the Assessment of Dental Caries Using Colour Dental Photographs.

D1ental caries remains the most common chronic disease in childhood, affecting almost half of all children globally. Dental care and examination of children living in remote and rural areas is an ongoing challenge that has been compounded by COVID. The development of a validated system with the capacity to screen large numbers of children with some degree of automation has the potential to facilitate remote dental screening at low costs. In this study, we aim to develop and validate a deep learning system for the assessment of dental caries using color dental photos. Three state-of-the-art deep learning networks namely VGG16, ResNet-50 and Inception-v3 were adopted in the context. A total of 1020 child dental photos were used to train and validate the system. We achieved an accuracy of 79% with precision and recall respectively 95% and 75% in classifying 'caries' versus 'sound' with inception-v3.

Evaluation of a deep learning system for automatic detection of proximal surface dental caries on bitewing radiographs.

OBJECTIVE: This study aimed to evaluate a deep learning (DL) system using convolutional neural networks (CNNs) for automatic detection of caries on bitewing radiographs. STUDY DESIGN: In total, 2468 bitewings were labeled by 3 dentists to create the reference standard. Of these images, 1257 had caries and 1211 were sound. The Faster region-based CNN was applied to detect the regions of interest (ROIs) with potential lesions. A total of 13,246 ROIs were generated from all 'sound' images, and 50% of 'caries' images (selected randomly) were used to train the ROI detection module. The remaining 50% of 'caries' images were used to validate the ROI detection module. Caries detection was then performed using Inception-ResNet-v2. A set of 3297 'caries' and 5321 'sound' ROIs cropped from the 2468 images was used to train and validate the caries detection module. Data sets were randomly divided into training (90%) and validation (10%) data sets. Recall, precision, specificity, accuracy, and F1 score were used as metrics to assess performance. RESULTS: The caries detection module achieved recall, precision, specificity, accuracy, and F1 scores of 0.89, 0.86, 0.86, 0.87, and 0.87, respectively. CONCLUSIONS: The proposed DL system demonstrated promising performance for detecting proximal surface caries on bitewings.

Deep learning for automated detection and numbering of permanent teeth on panoramic images.

OBJECTIVE: This study aimed to evaluate an automated detection system to detect and classify permanent teeth on orthopantomogram (OPG) images using convolutional neural networks (CNNs). METHODS: In total, 591 digital OPGs were collected from patients older than 18 years. Three qualified dentists performed individual teeth labelling on images to generate the ground truth annotations. A three-step procedure, relying upon CNNs, was proposed for automated detection and classification of teeth. Firstly, U-Net, a type of CNN, performed preliminary segmentation of tooth regions or detecting regions of interest (ROIs) on panoramic images. Secondly, the Faster R-CNN, an advanced object detection architecture, identified each tooth within the ROI determined by the U-Net. Thirdly, VGG-16 architecture classified each tooth into 32 categories, and a tooth number was assigned. A total of 17,135 teeth cropped from 591 radiographs were used to train and validate the tooth detection and tooth numbering modules. 90% of OPG images were used for training, and the remaining 10% were used for validation. 10-folds cross-validation was performed for measuring the performance. The intersection over union (IoU), F1 score, precision, and recall (i.e. sensitivity) were used as metrics to evaluate the performance of resultant CNNs. RESULTS: The ROI detection module had an IoU of 0.70. The tooth detection module achieved a recall of 0.99 and a precision of 0.99. The tooth numbering module had a recall, precision and F1 score of 0.98. CONCLUSION: The resultant automated method achieved high performance for automated tooth detection and numbering from OPG images. Deep learning can be helpful in the automatic filing of dental charts in general dentistry and forensic medicine.

Mobile photographic screening for dental caries in children: Diagnostic performance compared to unaided visual dental examination.

OBJECTIVES: This study was conducted to compare the use of intraoral photographs with the unaided visual dental examination as a means of dental caries detection in children. METHODS: Children aged 4- to 14-year-olds were visually examined at their schools. Following dental examinations, children had five photographs of their teeth taken using a smartphone camera. Four dental reviewers, who are different from those who visually examined the children, assessed intraoral photographs for dental caries. Sensitivity, specificity, and inter-rater reliability agreement were estimated to assess the diagnostic performance of the photographic method relative to the benchmark visual dental assessments. Caries prevalence was measured using dft/DFT (decayed and filled teeth) index. RESULTS: One hundred thirty-eight children (67 male and 71 female) were enrolled and had a mean age of 7.8 ± 2.1 years. The caries prevalence (dft/DFT > 0) using photographic dental assessments ranged from 30 percent to 39 percent but was not significantly different from the prevalence (42 percent) estimated with the visual dental examination (P ≥ 0.07). The sensitivity and specificity of the photographic method for detection of dental caries compared to visual dental assessments were 58-80 percent and 99.7-99.9 percent, respectively. The sensitivity for the photographic assessments was high in the primary dentition (63-82 percent) and children ≤7-year-olds (67-78 percent). The inter-rater reliability for the photographic assessment versus the benchmark ranged from substantial to almost perfect agreement (Kappa = 0.72-0.87). CONCLUSIONS: The photographic approach to dental screening, used within the framework of its limitations, yielded an acceptable diagnostic level of caries detection, particularly in younger children with primary dentition.

Deep learning segmentation of hyperautofluorescent fleck lesions in Stargardt disease.

Stargardt disease is one of the most common forms of inherited retinal disease and leads to permanent vision loss. A diagnostic feature of the disease is retinal flecks, which appear hyperautofluorescent in fundus autofluorescence (FAF) imaging. The size and number of these flecks increase with disease progression. Manual segmentation of flecks allows monitoring of disease, but is time-consuming. Herein, we have developed and validated a deep learning approach for segmenting these Stargardt flecks (1750 training and 100 validation FAF patches from 37 eyes with Stargardt disease). Testing was done in 10 separate Stargardt FAF images and we observed a good overall agreement between manual and deep learning in both fleck count and fleck area. Longitudinal data were available in both eyes from 6 patients (average total follow-up time 4.2 years), with both manual and deep learning segmentation performed on all (n = 82) images. Both methods detected a similar upward trend in fleck number and area over time. In conclusion, we demonstrated the feasibility of utilizing deep learning to segment and quantify FAF lesions, laying the foundation for future studies using fleck parameters as a trial endpoint.