Accuracy of Remote Diagnoses Using Intraoral Scans Captured in Approximate True Color: A Validation Study in Teledentistry

Background: Intraoral scans provide three-dimensional images with approximate true colors, representing possible tool in teledentistry for remote screening. Aim: To evaluate the levels of agreement between scorings derived from intraoral scans (IOS) and scorings derived from clinical examinations for assessing dental and periodontal conditions. Material and Methods: The sample included 10 patients in need of a restorative treatment. Following an intraoral scan, a full-mouth dental and periodontal examination was performed and periapical radiographs were taken. The intraoral scans were transmitted from the intraoral scanner to a cloud server. Ten examiners were asked to perform dental and periodontal assessments solely based on the intraoral scans. The scans were presented to the examiners on a tablet computer using the IOS-specic viewer app. Thereafter, the periapical radiographs were added and the examiners asked to do provide a second round of assessments. The remote scoring was compared to the clinical scoring derived from the clinical examination. The time for the remote assessment was recorded. The agreement between remote and clinical scorings was then analyzed descriptively. Results: Conclusion: The remote screening was effective in detecting patients in need for a clinical examination and was time-ecient. Further improvement of the image quality of intraoral scans may allow increasing the accuracy of remote assessments.


Introduction
Telemedicine is the remote delivery of health care services using information and communication technologies. The broad goal of telemedicine is to improve health of individuals and their communities in underserved areas (WHO, 1998). Teledentistry is a sub-group of telemedicine and is successfully used within the dental practice for teleconsultation, telediagnosis and delivery of oral care services (Khan, Omar, 2013). A systematic review concluded that teledentistry was a valuable tool for oral screening similar to face-to-face consultations (Alabdullah, Daniel, 2018). Several electronic devices were used to capture the intraoral soft and hard tissues such as smartphones, digital extraoral cameras, or intraoral cameras.
Smartphones equipped with imaging technology are readily accessible and very easy to use (Daniel, Kumar, 2014). In dental traumatology a clinical study showed that the remote diagnosis of dental traumas based on mobile phone pictures was similar to the diagnoses conducted in person (de Almeida Geraldino, Rezende, da-Silva, Almeida, 2017). A cross-sectional study showed, that smartphones with photo messaging can serve as an effective tool for the remote screening of potentially malignant oral disorders (Vinayagamoorthy, Acharya, Kumar, Pentapati, Acharya, 2019). Standardized mobile phone pictures were remotely assessed for dental caries without radiographs in children with mixed dentition (AlShaya, Assery, Pani, 2020). Sensitivity and speci city among several dentists were above 80%. A greater reliability was found at primary teeth as compared to permanent teeth. Similarly, a study showed that occlusal caries can be detected with acceptable diagnostic accuracy based on photographs taken by a smartphone camera compared to face-to-face screenings (Estai, et al., 2017). In orthodontics, the monitoring of linear tooth movements by means of a smartphone software showed an accurate assessment of the real tooth movements (Moylan, Carrico, Lindauer, Tufekci, 2019).
The use of small hand-held intraoral cameras was suggested as a feasible and potentially cost-effective option to a visual oral examination for caries screening in children (Kopycka-Kedzierawski, Billings, McConnochie, 2007). A clinical study with a total of 62 children reported that the agreement between digital and conventional clinical examinations was very good for various oral conditions (Pentapati, et al., 2017). Sensitivity and speci city were 98.1% and 66.7% for caries. Similar values were obtained for the detection of stains (99% sensitivity; 77.8% speci city), calculus (98%; 72.7%), and tooth wear (90.3%; 81%). Another study reported that sensitivity and speci city were 73% and 98% for the remote screening of dental caries in young adults by means of ve standardized intraoral photographs (Morosini Ide, de Oliveira, Ferreira Fde, Fraiz, Torres-Pereira, 2014).
To the best of our knowledge, no clinical study has evaluated the validity of using intraoral scans with approximate true colors for the screening for diseases affecting the periodontal tissues and dental hard tissues. Evidence that IOS are valid for the screening may change the organizational structure of dental practices leading to a higher adoption and a wider distribution of this technology. Therefore, the aim of this clinical validation study was to assess the agreement between remote screening based on intraoral approximate true color scans and traditional clinical examinations.

Material And Methods
The study protocol was approved by the Ethical Committee of the University of Zurich, Switzerland, and categorized as a study not being regulated by the law on human research in Switzerland (BASEC-Nr. Req-2019-01277). Signed informed consent was obtained from the patients included in this study.
Patients in need of a prosthetic rehabilitation were consecutively screened at the Clinic of Reconstructive Dentistry of the University of Zurich, Switzerland, and recruited for the present study. The patients were examined by experienced dentists, who took the intraoral scans and recorded the intraoral ndings.
An intraoral scan (Trios, 3Shape, Copenhagen) with approximate true colors was obtained in all patients.
The scan included a full-arch scan of both jaws including all teeth and a bite registration of both sides in occlusion. The computed scans were issued with a separate study number to pseudonymize the intraoral scans before being sent to the store-and-forward based server (portal.3shapecommunicate.com).
Detailed dental records were obtained including number and location of teeth, location of caries lesions, of non-carious cervical lesions (NCCL), and of erosive lesions. The location, size, and material of llings and restorations were also recorded. In addition, tooth mobility (Flemming, 1993) and tooth vitality by means of CO2 testing were assessed. A full-mouth periodontal examination (FMPE) was recorded including probing depth, attachment levels, plaque index (Silness, Loe, 1964), bleeding on probing (Ainamo, Bay, 1976), furcation involvement, and gingival recessions. Periapical radiographs were taken when clinically indicated.
The clinical data for each patient were captured in a standardized assessment form and served as control (Table 1). In brief, the amounts of plaque and calculus were rated using a scale with 5 scores, whereas dichotomized scores were applied for the parameters gingival recessions, furcation involvement, toot erosion, tooth wear, stains and NCCL. The existence of decayed, lled, and crowned teeth as wells as the presence of implants was checked. In addition, the quality of the restorations was evaluated. Gingival and periodontal health were evaluated using a scale with 3 scores (healthy, localized, generalized).
For this validation study, ten patients were selected. The sample of patients was created to represent a wide variety of different clinical situations (Fig. 1). Edentulous patients and patients presenting with Kennedy class 1 were excluded. The overview of the clinical characteristics of the ten patients is presented in Table 2.
The remote examiners, who had a minimum of three years of experience as practicing dentists, assessed the intraoral scans on a tablet computer (iPad Air, model A1474, Apple Corp. Cupertino, CA) using the software app provided by the manufacturer (3Shape Communicate, Version 4.0.1). The examiners used the standardized assessment form during the tele examination. Thereafter, they were provided with the periapical radiographs and asked to provide a second round of tele assessment. The time for the tele examination was recorded for assessment of the intraoral scans as well as for the additional assessment with the radiographs. The remote scoring (test) was compared to the scoring derived from the clinical data (control).
Each of the 10 patients was scored remotely by 10 examiners resulting in a total of 100 observations per parameter. Data was coded in Microsoft Excel. Analyses and illustrations were generated with the statistical software R (R Core Team, 2019), including the package ggplot2 (Wickham, 2016). A descriptive sub-analysis was performed for the parameters plaque and calculus comparing the remote scorings from dentists with less than 5 years of clinical experience with scorings from dentists with more than 5 years of clinical experience.

Results
The sample of examiners comprised a total of 10 clinicians aged between 28 and 41 years (4 females, 6 males). None of the examiners has ever done a remote examination by means of approximate true color scans. The mean time (± standard deviation) for the tele screening was 3.17 (± 1.15) minutes and the additional consultation of the radiographs accounted for another 1.48 (± 0.59) minutes.
The agreement for the dichotomized clinical indices (gingival recessions, furcation involvement, toot erosion, tooth wear, stains and NCCL) ranged between 78 and 95%. The presence or absence of decayed teeth was correctly assessed in 73% of the cases. This accuracy was higher for lled teeth (98%), crowned teeth (97%), and implants (96%). The provision of the radiographs allowed the remote examiners to correctly detect all lled teeth, crowned teeth, and implants, whereas the detection of decayed teeth (70%) was not improved. The false negative rate for the detection of decayed teeth was 21% and 23% with the provision of radiographs. The false positive rate increased from 3% without radiographs to 23% with radiographs.
The remote screening of the dentition for the presence or absence of insu cient llings and crowns was correct in 63% and 65% of the cases and the provision of radiographs allowed to improve the accuracy for crowns (78%), but not for llings (65%) (Fig. 2). The evaluation of the llings resulted in 29% false negatives and in 7% false positives (31% and 4% with radiographs). The assessment of crown quality revealed 21% false negatives and 15% false positives with a decrease when radiographs were consulted (16% false negatives and 6% false positives).
The perfect agreement between the remote assessment and the clinical examination was 35% for plaque and 40% for calculus (Fig. 3). The tele consultation resulted in 35% higher plaque score values (overrating) and in 30% lower plaque score values (underrating) as compared to the clinical examination. The presence of calculus was overrated in 10% of the observations, whereas underrating encompassed 51%. A moderate agreement represented by a score deviation of ± 1 unit was found for 48% of the plaque scores and 50% of the calculus scores, whereas a low agreement with score deviations of ≥ 2 units was limited to 17% (plaque score) and 11% (calculus score). The visual representation based on the clinical experience did not reveal a trend (Fig. 3).
The sensitivity and speci city values were 0.61 and 0.39 for gingivitis and 0.67 and 0.33 for periodontitis, with no relevant changes when radiographs were provided for the diagnosis of periodontitis (0.72 and 0.28) ( Table 3). The false negative rate for gingivitis and periodontitis was 11% and 3%, respectively. The false positive rate was 28% for gingivitis and 30% for periodontitis.

Discussion
The results of the present study showed that a dentist can make a time-e cient and valid tele screening of the oral soft and hard tissues based solely on approximate true color intraoral scans taken by means of IOS. The agreement between remote diagnosis and clinical diagnosis for dental and periodontal assessments, however, was low.
The continuous development of hard-and software has made it possible to electronically capture and transmit clinical pictures of patients. The accuracy of teledentistry, however, depends on the selected eld of view. Two-dimensional intraoral images have their shortcomings, because they represent a limited twodimensional view of three-dimensional soft and hard tissue structures. In a clinical study on the e cacy of remote screening for dental caries, up to 15.4% of all screened teeth could not be scored because of missing image data (Estai, et al., 2016)). In the present study, all teeth were accessible for the remote screening because intraoral scans provide the full three-dimensional structure of teeth and gingiva. The fact that calculus scores were underrated, however, con rmed that only structures represented on the images can be remotely assessed.
In addition, the quality of the images may in uence the outcome of tele evaluations. A clinical study providing smart phone pictures for the remote screening of potentially malignant disorders demonstrated that the false negative rate decreased as the camera resolution increased (Haron, et al., 2017). The results of the present study showed that the assessment of the remote score for the amount of plaque and calculus was limited. This may be related to the quality of the color images that the software uses for the colorimetric augmentation of the three-dimensional le. Anecdotal comments by the remote examiners suggested that the image quality on the tablet software may have been lower than in the IOS software emphasizing that hard-and software may in uence the accuracy of remote screening.
The requirements of valid screening tools are time e ciency and accuracy. It was reported that the establishment of a FMPE took a mean of 29 minutes (Owens, Dowsett, Eckert, Zero, Kowolik, 2003) and up to 40 minutes (Benigeri, Brodeur, Payette, Charbonneau, Ismail, 2000). Therefore, tests and indexes were established to estimate the prevalence of periodontitis. A systematic review reported sensitivity values of partial-mouth examination protocols ranging from 57-96% with a higher sensitivity in protocols including the examination of all teeth (Tran, et al., 2013). The remote assessment of approximate true color scans showing the gingival structures resulted in a sensitivity of 67% for the prevalence of periodontitis. More importantly, the false negative rate for periodontitis was very low (3%) demonstrating a low risk of missing out patients in need for a periodontal therapy.
A systematic review reported a range for sensitivity values by means of visual inspection for caries detection between 0.35 and 0.78 (Gimenez, et al., 2015). In the present study the remote inspection of intraoral scan data revealed a sensitivity of 0.88 for the detection of decayed teeth. The false negative rate for the detection of decayed teeth was high. Therefore, the present method cannot be recommended for caries screening. Recent IOS, however, are equipped with uorescent or near-infrared light transillumination technology to detect caries lesions concomitant with the acquisition of the threedimensional images. A clinical study showed the potential of this new technology (Kunisch, Schaefer, Pitchika, Garcia-Godoy, Hickel, 2019). Near-infrared light images allowed to detect proximal caries in posterior teeth similar to bitewing radiographs and were superior to the visual inspection alone.

Page 7/15
The recording time as well as the investment costs for an IOS have to be considered. General dental practitioners showed concerns in relation to excessive time spent in capturing images and transmission of information (Stephens, Cook, Mullings, 2002). The dental practices' organizational structure, however, may be adapted by delegating image acquisition to dental auxiliary staff. Today, IOS are primarily used for the fabrication of reconstructions and dental appliances. The potential advantages associated with the use of IOS for remote screening of early diagnosis and the need for treatments may justify the high investment costs.

Conclusion
The use of approximate true color intraoral three-dimensional images acquired by means of IOS can be a helpful tool for the screening of visible soft and hard tissue lesions.

Declarations
Ethics approval and consent to participate: Not applicable Consent for publication:

Not applicable
Availability of data and materials: All data generated or analysed during this study are included in this published article.

Competing interests:
The authors declare that they have no competing interests.