A proof-of-concept evaluation of a cloud-based store-and-forward telemedicine app for screening for oral diseases

Introduction

Despite significant progress in reducing oral diseases in developed countries over the past three decades, dental caries remains a significant public health problem, particularly amongst disadvantaged people and people living in remote areas. ¹ In most developed countries (including Australia), the majority of oral health professionals work in private practice and clustered in cities, and remote and regional areas suffer from a significant shortage of dental practitioners.^2,3 Delay in seeking dental care could be considered a major contributor to poor oral health. ⁴ This can happen not only in rural or remote areas where access is primarily limited by geography, but also in underserved urban regions where access is limited by a lack of socioeconomic resources or complexity of life. ⁵

Due to remoteness, shortage of dentists, lack of insurance or financial burden, in Australia only half of all adults visit a dentist at least once a year. ⁶ Thus, it is essential to seek strategies that can increase access to adequate care and avoid long waiting without compromising quality, effectiveness and safety. One of a number of growing solutions to address limited access to dental care is the utilization of telemedicine technologies in the screening for oral diseases, providing care, evaluation of care and referral. There are two telemedicine modalities: real-time consultation and store-and-forward. ⁷ For most dental applications, the store-and-forward method provides excellent results without excessive costs for equipment or connectivity. ⁸ While email remains a popular method for store-and-forward consultation with limited data security, videoconferencing is still a popular method for real-time consultation. ⁷ However, video conferencing has its limitations as it needs expensive equipment and requires all parties to be online at the same time. Cloud storage represents another store-and-forward method that allows storing and retrieving data through a secured system. ⁷ Despite adoption rates for email and videoconferencing within the field of dentistry increasing rapidly, adoption of cloud computing is still lagging. ⁷

With smartphone camera technology improving significantly, utilization of smartphones to obtain and transmit images for screening of dental caries has recently matured.^9–11 Similar to dedicated digital oral cameras, smartphone cameras have zoom and flash features, as well as manual adjustments that allow easier capturing of intraoral or extra-oral images. The portability and accessibility features of smartphones can provide an effective means for capturing images in less time and be less intimidating for children. ⁷ Recent studies reported that children were enthusiastic and cooperative during photography and recording procedures with smartphones. ¹² The process of capturing images using smartphones can therefore be noninvasive and less stressful for small children than the usual oral examination. ¹² This study was designed to develop a suite of technology harnessing the advantages of ubiquitous smartphone imaging technologies and cloud-based computing and to build an easy to use, accurate tool to enhance the opportunity for dental screens for patients with otherwise limited access to dental care.

Methods

System architecture

Based on a telemedicine platform ‘Remote-i’, which was developed for ophthalmology and other medical applications, the Australian e-Health Research Centre (AEHRC) has built a store-and-forward based teledentistry system to carry out a screening programme for dental caries. Remote-i is a comprehensive data management server that has been widely used as a telemedicine platform in various screening programmes in Australia and China. ¹³ The system enables uploading and storing images online either directly from a smartphone or from a computer. In addition, it is capable of image acquisition, data entry, storage and retrieval of data through a password-protected system (Figure 1).

OPEN IN VIEWER

Figure 1.

Overall architecture and workflow of the teledentistry system.

Smartphone application (Android)

An image acquisition Android application was developed and installed on the Motorola® Moto smartphone (USA) to facilitate entering patient personal details and capturing dental photos, and then uploading data corresponding to each patient to cloud storage, using Wi-Fi hotspots or the cellular data networks (Figure 2). At this stage, the app only supports Android devices, but an app that can support other operating systems such as IOS and Windows is under consideration. The patients’ data stored on the database system are accessible via a specially designed web interface. The web interface also allows users to plug in any USB dental camera, capture images and add them to the current patient’s record without the need to import them from the mobile device. The research team can also access the server and follow up with the remote screening site for further action (Figure 1).

OPEN IN VIEWER

Figure 2.

Illustrating the teledental Android phone application used for dental photography.

Proof-of-concept trial protocol

All data collection was completed under ethics approval from The University of Western Australia. Six adult volunteer patients enrolled in the proof-of-concept trial. Each participant received an in-person oral examination by a registered dental practitioner (without radiographs) to record caries and existing restorations; this was used as the gold standard. The onsite dentist spent an average 10 minutes per patient for oral screening. The dentist recorded data on an oral assessment sheet aligned with World Health Organization (WHO) protocol ¹⁴ and provided treatment recommendations or referral (Figure 3). In a separate visit, a trained teledental assistant took photographs of each participant’s oral cavity using a smartphone application. Techniques for obtaining each oral view were demonstrated during training, and each teledental assistant is provided with image photography protocol. Cheek retractors and intraoral mirrors were not used in the screening so as to simulate the final expected working conditions. An average of five oral images were taken per patient including: one anterior view; one upper occlusal view; one lower occlusal view and two lateral view images. Additional views of the teeth, intraoral soft tissue or extra-oral structures were taken in some cases. The photographs were then directly transmitted from the smartphone to a secure online server (Remote-i) for evaluation by an offsite grader (Figure 3). Although the patient’s data transmitted to the server is de-identified or anonymous, it was not de-identified to the main investigator or clinicians as we used patients ID numbers in the database as a key. Techniques for obtaining each oral view were demonstrated during training, and each teledental assistant was provided with image photography protocol.

OPEN IN VIEWER

Figure 3.

Figure 3(a) shows a copy of the onsite oral health assessment form. Figure 3(b) depicts an online dental chart for a patient, which includes sections for personal patient details, oral images and the oral assessment form.

Findings from the proof-of-concept trial

A complete set of 30 oral images was obtained from all six participants using the smartphone’s camera and uploaded to the Remote-i server (Figure 3). All uploaded files were stored in a database in an uncompressed JPEG format; the average file size was 900 KB. Participants’ characteristics collected during the trial are summarized in Table 1. It took approximately one minute to register a patient (entering personal details) and five minutes to take pictures. The quality of images obtained by the teledental assistants was assessed by an independent practitioner to ensure that they were complete, colour-balanced, focused and clear. Oral images of poor diagnostic quality were re-captured when necessary.

OPEN IN VIEWER

Table 1.

Participants’ characteristics.

	Age (in years)	Sex	Indigenous status
Participant 1	33	Male	None
Participant 2	61	Male	None
Participant 3	53	Male	None
Participant 4	22	Male	None
Participant 5	47	Female	Yes
Participant 6	22	Male	None

The proportion of the captured digital images that were gradable was 100%. Out of 192 teeth reviewed, the proportion of ungradable teeth was 8%. Statistical testing of the reliability of the two sets of oral examinations (clinical and teledental) and two graders (grader 1 and grader 2) was completed using kappa statistics. The calculation of sensitivity and specificity of teledental screening was based on the presence or absence of untreated caries; the sensitivity of the teledentistry examination was 57% and specificity was 100% versus in-person oral examination (Table 2). The inter-grader agreement between traditional oral examinations and teledentistry examinations was 70%, while inter-grader agreement between teledental grader 1 and teledental grader 2 was 62% (Table 2).

OPEN IN VIEWER

Table 2.

Accuracy and reliability of teledental screening.

Comparison of oral health assessments
	Sensitivity (%)	Specificity (%)	Kappa
Gold standard vs teledental	57%	100%	0.70
1st grader vs 2nd grader	60%	99%	0.62

Discussion

Our findings suggested that the combination of a smartphone camera and store-and-forward technology for oral health screening can be adequate for screening purposes and offers a reliable alternative to traditional oral examinations. The trial showed at the level of proof-of-concept that there was a good concordance between teledental and face-to-face oral assessments. This reflects previous research which shows no statistical difference between teledentistry and clinical screening for caries, with Kappa statistics for inter-grader reliability between clinical and teledentistry screenings ranging from moderate to very good agreement.^15–18

Telemedicine has become a viable option to address limited access to specialist care, particularly in locations far from specialist centres. Specialists do not need to be physically present with the patient, as they can provide consultation and guidance for local practitioners at a distance. Teledentistry would expedite early diagnosis and facilitate timely treatment of oral diseases by reducing the isolation of practitioners and increasing access to specialist care. ² In addition, it has the potential to improve the quality of care for the underserved by facilitating the provision of timely information to dentists for better decision making and effectively triaging patients who require specialist consultation, thus reducing waiting lists and supporting locally based oral treatment. ¹⁹ Telemedicine applications within dentistry are highly diverse, including preventive dentistry, oral medicine, oral and maxillofacial surgery, orthodontics, periodontics and education, with preliminary evidence suggesting positive outcomes in comparison to clinical settings. ²⁰

The quality of images and the capability to grade images accurately are very important factors when evaluating the feasibility of telediagnosis of diseases. ²¹ During testing we found that many of the oral images were blurred or over saturated when the default camera application was used. This is because the flash of the camera is too bright, and when we move the camera close to the mouth it will overexpose the area and the resulting image will be bleached out. However, a third party camera app has a feature called ‘the torch mode’ (e.g. Camera360 Ultimate, China), which can keep the flashlight turned on during the initial focus and will not change until the capturing is finished. Using this mode, we were able to obtain high quality photos with proper focus and illumination. This is likely due to the ability of the mobile phone computer to analyse the over-exposed areas and adjust its exposure values appropriately.

The smartphone camera used in this study produces images of 10 megapixels, and is considered adequate for producing good-quality images. However, in some cases there was concern about the loss of detailed diagnostic information due to the poor quality of the images. Because graders sometimes found difficulty identifying caries spots and differentiating them from darkened defects or artefacts, some teeth were graded as unrated or not shown by teledental graders. Depending on the lighting, identifying amalgams or composites was also a challenge. Although training teledental assistants to obtain good quality oral images did not take much time, adherence to the given photography protocol was a challenge especially in the situation where patients are uncooperative or stressed, such as preschool children. In this trial, the sensitivity of the teledentistry examination was 57%, and specificity was 100%. The sensitivity value of 57% was because missing and filled teeth were not included in the analysis, as they can be easily recognized on digital images. In addition, sub-optimal sensitivity and specificity ratings are often associated with poor quality images which are likely attributed to the method, limitations of digital image acquisition (e.g. lighting, autofocus) or non-optimal image review modalities (e.g. monitors). ²¹

The shortcomings of current policy that govern telemedicine services have been a matter of concern for many practitioners. ²² Lack of remuneration, licensure and taxation guidelines as well as medico-legal and copyright issues have created a barrier to the adoption of teledentistry in practice. ²² Well-defined and efficient policies and laws are essential to regulate teledentistry practice and expand the use of technology in clinical practice. Despite teledentistry being applicable in both public and private dental settings, barriers impeding the integration of telemedicine technology into dental practice exist. The increasing costs of setting up telemedicine equipment were regarded by many advocates of telemedicine as a major obstacle. ²⁰ However, almost all dental practices possess intraoral cameras, digital radiographs, computers and internet access, which provide the core infrastructure for teledentistry practice. ²³ The use of available infrastructure can reduce overall practice expenses and consultation fees. Integration of teledentistry into practice has the potential to address certain aspects of routine dental practice such as reducing consultation times and practice expenses, as well as improving practice management, referrals and patient satisfaction. ²⁴ Further well-designed efficacy and cost-effectiveness trials in rural or remote communities are needed to determine whether specific populations may benefit from the use of teledentistry.

Conclusion

In order to improve oral health for the underserved or people living in rural and remote areas we need robust and readily deployable screening tools. Simple telemedicine applications hold considerable promise in bridging the gap in oral healthcare for the remote or rural communities. This trial shows that teledental screening has the potential to be utilized as a valid and reliable screening tool to identify high-risk individuals with decay and can allow onsite practitioners to triage referrals in a timely manner and treat more patients. However, this study was only a proof-of-concept trial and a full study is needed to confirm the accuracy and reliability of the teledentistry system.