Diagnostic accuracy of intraoral mobile photography for oral health screening in children: a pilot study

Abstract

Background

See More

Mobile dental photography is increasing in relevance in the diagnosis of oral diseases. This study aimed to evaluate the diagnostic accuracy of intraoral mobile photography in assessing caries experience intensity (decayed, missing/extracted, filled teeth [DMFT/deft]), simplified oral hygiene index, and modified gingival index.

Methods

This study included 358 children of 7–12 years of age. A clinician evaluated DMFT/deft; simplified oral hygiene; and modified gingival indices by visual examination. Simultaneously, dental students recorded intraoral photographs with a mobile phone in eight predetermined projections. Another oral professional calculated indices based on the photographs. Sensitivity, specificity, and positive and negative predictive values of dental photography were evaluated, and Cohen’s kappa was calculated.

Results

A total of 2864 photographs were evaluated in this study. The sensitivity and specificity of mobile photography for DMFT; deft; simplified oral hygiene, and modified gingival indices were 95.8 (95% confidence interval: 93.4–98.3) and 89.2 (83–95.5); 100.0 (100–100) and 88.6 (79.3–98); 89.7 (85.8–93.5) and 91.4 (86.3–96.5); and 77.6 (67.6–87.6) and 93.8 (91–96.6), respectively; positive predictive values were 96.2, 97.7, 95.6, and 74.3; negative predictive values 88.3, 100.0, 80.9, and 94.8; and Cohen’s Kappa values were 0.848, 0.928, 0.784, and 0.702, respectively.

Conclusions

Intraoral mobile photography, based on appropriate guidelines, is reliable for assessing DMFT/deft; simplified oral hygiene; and modified gingival indices in children and could be a useful tool in dental public health, supporting the involvement of dental students in similar research studies.

Background

Intraoral dental photography is considered a practical way of expressing, sharing, storing, and transmitting information. In 1985, Wolfgang Bengel created standards that are still relevant today [1]. Modern literature also mentions that if dental photography is subjected to rules and specific guidelines are developed, it can be called scientific photography and used in scientific research [2, 3, 4]. The use of dental photography for screening in dental public health has become more relevant in recent years due to rapid development, accessibility, and increased demand for digital technology and telemedicine [5, 6, 7].

In dental practice, the use of professional cameras and flashes is an indispensable tool for relatively small-scale clinical studies and preparation of case reports [8, 9]. The recent technology evolution has led to the advancement of mobile intraoral photography and the introduction of its capabilities into dental practice. The latest generation of mobile phones are equipped with sensitive cameras, powerful lighting, and large capacity, which allows to create a large number of high-quality and high-resolution photographs during the research process. Mobile phones considerably simplify the process [10]. Teledentistry has high potential, and more studies are needed to further substantiate the utility of this new technology in epidemiological studies [11, 12]. Development of dental photography is necessary for the sustainable access to dental services [13]. Despite the increased use of dental photography during the coronavirus disease (COVID-19) pandemic, research has shown a need for higher staff training and promotion to ensure its integration and effective use among dental professionals [14].

Studies have reported on researchers trying to create applications based on dental self-recorded photographs to monitor caries [15, 16], gingivitis [17, 18], and oral mucosal lesions [19] to develop user-centric monitoring processes. The relevance of intraoral mobile photographs recorded by non-professionals in large-scale screening studies for dental public health and whether the technological evolution of mobile cameras allows this is unclear with limited evidence. The scientific confirmation of this issue aims to adapt dental screening processes to important requirements such as saving time and reducing the resources of medical inventory and oral professionals for screening, easy communication, and accessibility.

Therefore, this study aimed to evaluate the diagnostic accuracy of intraoral mobile photography in assessing caries experience intensity (decayed, missing/extracted, and filled teeth [DMFT/deft]), simplified oral hygiene index (S-OHI), and modified gingival index (MGI) in children of 7–12 years of age. To achieve this goal, oral health indicators, such as DMFT index for caries experience intensity in permanent teeth, deft index for caries experience intensity in primary teeth, S-OHI, and MGI, were assessed using visual screening and compared with same indicators assessed using mobile intraoral photographs.

Methods

We used the Standards for Reporting of Diagnostic Accuracy guidelines for greater accuracy of the research method [20].

Population selection and inclusion criteria for the study

The study involved 358 school-attending children of 7–12 years of age. Study participants were selected from another population-based research (n = 421) that was conducted to evaluate the oral health of post-COVID children in Tbilisi (Georgia) during 2022–2023. All of them were offered to simultaneously participate in the study to evaluate the diagnostic accuracy of intraoral mobile photography following the criteria developed.

The criteria for inclusion in the study were as follows: age of 7–12 years; the child’s consent to a visual examination of the oral cavity, intraoral photography of teeth, and staining of dental plaque; and written consent of the parents for the inclusion of the child in the study.

The exclusion criteria were as follows: incorrect photograph protocol (dental arch is not completely visible, blurry photograph, or no projection required by the standard); insufficient staining of the plaque (the child could not rinse the solution for staining the plaque); fixed orthodontic appliances; refusal of the child to participate in the study, record an intraoral photograph, or rinse the disclosing agent in the oral cavity; and parents’ refusal to include their child in the study (Fig. 1).

Fig. 1

Flowchart showing participant selection in the study. DMFT, decayed, missing, filled teeth; deft, decayed, extracted, filled teeth; S-OHI, simplified oral hygiene index; MGI, modified gingival index

Assessment of oral health indicators (variables)

Oral health indicators were assessed according to the World Health Organization (WHO) standards recommended for massive and epidemiological studies [17, 18, 21, 22] (Supplementary Table S1 in Additional File 2). All three indices were used with both continuous and binary values, such as DMFT/deft = 0 or ≥ 1 for caries experience intensity, implying the absence or presence of carious lesion; S-OHI ≤ 1.2 or > 1.2 for oral hygiene assessment, where a value of up to 1.2 was considered good hygiene and a value > 1.2 was considered absence of good hygiene; and MGI = 0 or ≥ 1 for gingival lesion, implying the absence or presence of gingival lesions.

Visual screening

Visual examination of children was conducted in school medical rooms. The children were instructed to sit on a chair by the window. Natural light was used. The doctor used disposable dental mirrors. The data for calculation of DMFT/deft and MGI were dictated by the doctor to an assistant, who recorded them in the patient’s medical card. To assess S-OHI, the children rinsed the erythrosine-containing disclosing agent, which does not contain alcohol and is safe, in their mouths for 30 s, after which the doctor evaluated the S-OHI and the assistant recorded the data in the patient’s medical card.

Intraoral mobile photography

Based on general experience in intraoral dental photography [1, 23], we standardized the mobile intraoral photography process to ensure that the information obtained by photographing of beneficiaries involved in the screening process was reproducible, equally informative, and relevant to all three indices assessed in our research. All photographs were recorded with one phone, i-Phone 11 (Model A2111; designed by Apple in California and assembled in China); participants were sitting on chairs in the same position and same lighting and mobile flash light was used; no filter was used, shooting distance was at 9–10 cm from the lips, zoom 1.8–2.1, phone was positioned horizontally, and eight predefined projections were recorded for all participants. The first three projections were recorded before plaque staining and the next five after staining (Supplementary Figures S1–S8 in Additional File 1). The first three projections were used to calculate the DMFT/deft and MGI indices. The next five projections were used to calculate the S-OHI index and to confirm the DMFT, deft, and MGI indices.

Two students of the dentistry program recording intraoral dental photographs were trained in advance and studied the research standards and criteria. During photography, the main researcher, who was a clinical doctor did not help the students. To protect the privacy of the research participants, students were not allowed to take commemorative photographs with the beneficiaries or to have their faces recorded in the photographs. The camera settings used were 12 MP with 120° ultra-wide (f/2.4) aperture, 2× optical zoom, and up to 5× digital zoom with true tone flash.

Storage of participant data

The examined children were assigned numbers, and their medical records and photographs were numbered accordingly. The photographs recorded by the students were saved in individual folders with the corresponding number. Photographs were copyrighted with the specific marking and forwarded to a second independent oral professional researcher. The data were analyzed using statistical software SPSS version 23 (IBM Corp. Released 2015. IBM SPSS Statistics for Windows, Version 23.0. Armonk, NY).

Statistical analysis

Students presented 4017 photographs, from which 2864 photographs were used. The values obtained by visual intraoral examination were considered as the reference standard. Sensitivity, specificity, and positive and negative predictive values (PPV and NPV, respectively) ​​were calculated according to the data in Table 1 and the provided formulas [24, 25, 26].

Table 1 Probability table to estimate sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV)

The reliability of inter-rater agreement was estimated by calculating the index of agreement between two independent researchers, Cohen’s kappa (κ). The minimum acceptable level of Cohen’s kappa was determined to be 0.61. Cohen’s kappa typically ranges from 0 to 1. Values from 0.6 to 0.8 indicate significant agreement and from 0.8 to 1.0 indicate almost perfect agreement [27].

All variables were used with binary values ​​to determine the sensitivity, specificity, PPV, NPV, and inter-rater reliability of intraoral mobile photography and 95% confidence intervals (CIs) were estimated. The variables were used as continuous variables to determine the mean of oral health indices and the Pearson’s correlation test was used to evaluate the correlation between oral health variables obtained by visual and photograph screenings.|r|>0.7 was defined as the minimum acceptable level of Pearson’s correlation and p ≤ 0.05 was considered as the level of significance.

Research ethics (study approval)

The ethical approval of the study was obtained from the Biomedical Research Ethical Council of the School of Health Sciences of the University of Georgia (research code UGREC − 04 − 22/ 09.03.2022). In accordance with the Declaration of Helsinki, informed consent was obtained from the guardians of the participants in the study prior to inclusion in the study. The parents of all examined children signed an informed consent document confirming that they do not object to the use of intraoral photographs in the article. Research data are maintained confidential. The Ministry of Education and Science of Georgia gave permission to conduct screening in schools (Doc code MES 9 22 0000871059).

Results

Descriptive statistics

This study included 358 (85%) of the 421 children selected from another population-based research. The parents’ response rate for participation in the study was 100% and children’s response rate was 92.17%. Of the 358 children, 171 (47.8%) were girls and 187 (52.2%) were boys with a mean age of 9.79 years (standard deviation [SD] = 1.6). Mixed and permanent dentitions were recorded in 261 (72.9%) and 97 (27.1%) participants, respectively. The total number of teeth studied was 8471, including 2033 primary and 6438 permanent teeth. The mean and SD of oral health indicators were determined based on the results of visual and photograph screenings and the obtained values differed slightly for each studied indicator (Table 2).

Table 2 Oral health indicators based on visual and photograph screenings (based on continuous variables)

Inferential statistics

Sensitivity exceeded 95% for both primary and permanent teeth. The specificity of caries experience intensity in primary and permanent teeth ranged from 88.6 to 89.2%. Component D/d had the highest sensitivity (94.1–94.6%). Relatively low sensitivity was observed in the M/e (83.3–73.7%) and F/f (90.7–77%) components. The PPV and NPV for DMFT and deft were high at ≥ 88.3%. Inter-rater reliability was higher for deft (κ = 0.928) than for DMFT (κ = 0.848) (Table 3).

Table 3 Diagnostic accuracy and inter-rater reliability of mobile photography for the studied oral health indicators

The sensitivity of the S-OHI was the lowest at 77.6% compared to that of other indices and the PPV was 74.3%, which implies that the probability of false-positive results for S-OHI was higher than that for other oral health indicators. The inter-rater reliability was 0.702, which is also the lowest compared to that of other measures. The specificity and NPV for S-OHI were 93.8% and 94.8%, respectively. Accordingly, these values ​​indicate that the likelihood of false-negative results for this indicator is minimal (Table 3). For MGI, the sensitivity was 89.7%, specificity 91.4%, the probability of false-positive results > 90%, and the probability of false-negative results 80.9% (Table 3).

To evaluate the results of the study, Pearson’s correlation was also used to determine the relationship between continuous variables obtained through visual and photograph screenings. A strong association was observed for all indicators of oral health with|r|>0.7, with the highest for DMFT (r = 0.95) and lowest for S-OHI (r = 0.76; p < 0.001). For clarification of the results, you can see the ROC analysis in the Additional File 3.

Discussion

This study confirms that intraoral photographs recorded by non-professional dental students with mobile phones following the predetermined guidelines are relevant for the assessment of DMFT/deft, MGI, and S-OHI indices in children during dental screening. The study showed that the mean of caries experience intensity in primary and permanent teeth by both visual and photograph screenings is similar and is perceived in the same WHO assessment category (2.7–4.4: average caries experience intensity), similar to the means of S-OHI and MGI. Calculating and comparing means using continuous variables is a statistical method to suggest that the potential of photographic screening is sufficient.

High sensitivity and specificity support the quality of mobile photography as a test. It is a diagnostic test indicator that is not influenced by the prevalence of the disease, although confounders may affect sensitivity and specificity [28]. The prevalence of caries and gingival lesions in the studied population can be considered as a confounder. Prognostic values ​​vary according to the prevalence of the disease in the population. Even with a highly specific diagnostic test, if the disease is rare in the people tested, most positive test results will be false positives with a low PPV. Thus, determining PPV and NPV allows us to identify and avoid the risk of false-positive and false-negative cases [24].

Among the oral health indicators studied in this research, a relatively low PPV value was revealed for the S-OHI (74.3%). NPV for all studied indicators was ≥ 88.3%. Therefore, mobile photography can be considered sensitive and specific.

According to the study, sensitivity of mobile intraoral photography for deft was 100% and for DMFT 95.8%. A similar tendency has been reported in another study, in which a similar age group was studied; the sensitivity of mobile photography in primary dentition was 82% and 78% at > 7 years of age [29]. Another study found that mobile photography of primary teeth had higher sensitivity, regardless of whether the photographs were recorded by dental professionals (95%) or non-professionals (98.3%) [28]. A pilot study from Malaysia examined enamel and dentin caries using a mobile phone, and the sensitivity was 74% and 72%, respectively, and specificity was 100% for both [30]. A study conducted in 2009 among children of 3–7 years of age reported the sensitivity of intraoral photography for primary teeth as 85.5% [31]. With the improved technologies of intraoral cameras and mobile smartphones in the last decade, the mean of sensitivity and specificity for caries diagnosis has reached 80% [32], which was 60% in 2017 [15].

On evaluation of each component of DMFT/deft index separately, the sensitivity of component d was higher (94.6%) than the sensitivity of component D (94.1%). However, contradictory results were observed in the sensitivity of the M/e and F/f components. On considering the M/e component, the sensitivity of mobile photography for the extracted primary teeth was lower (73.7%) than that for permanent teeth loss (83.3%). The difference may be because during the visual examination, the child’s age was known to the physician, but second examiner assessed the photographs blindly and could not identify prematurely removed carious teeth. Thus, the reliability of agreement between researchers was weakest for the e-component (κ = 0.59, CI: 0.45–0.72) and was not considered sufficient since it cannot exceed the minimum level of agreement (0.61). Mobile photographs are more sensitive to the F/f component of permanent teeth, which may be explained by their greater volume and visibility.

Among the studied variables, the S-OHI had the lowest sensitivity of 77.6%, which could be because the soft white plaque was relatively poorly perceived by the mobile phone camera. Cohen’s kappa was 0.702 (CI: 0.583–0.788), which correspond to the minimum level of the specified kappa indicator of 0.61. In a previous study, the oral hygiene index in preschool children showed very high reliability (intraclass correlation coefficient = 0.987) in diagnostics using mobile photography [33]. Further, the sensitivity of mobile photographs for debris index has been studied in children of 7–12 years of age, with a result of 96% inter-rater reliability, which is a stronger agreement than that achieved in our study; however, this study used a higher generation mobile phone (iPhone 13) [29].

The sensitivity of mobile photography for the MGI index was 89.7%, specificity 91.4%, PPV 95.6%, NPV 80.9%, and inter-rater reliability κ = 0.78. While previous studies have primarily focused on caries experience intensity, our study is among the first to rigorously assess the accuracy of mobile photography in evaluating both MGI and S-OHI. To our knowledge, the research directly comparing MGI assessments using mobile photography to clinical examinations is scarce, highlighting the innovative nature of our findings.

Pearson’s correlation was used to compare variables based on their absolute values. A strong correlation of r > 0.7 was obtained for all indicators. Among them, the lowest correlation was observed for S-OHI. The reliability of the research is confirmed by that the researchers previously standardized the photographic conditions according to Bengel’s study [1], thereby optimizing the photograph projections and adapting them to the study design.

The results of our research are similar to those of other studies that discuss the need for development of dental photography from the perspective of sustainable access to dental services [13]. However, this study differs from other studies in its scale and number of study participants, which increases the value of the study from an epidemiological aspect. Thus, the present study could provide a basis for epidemiological studies using mobile intraoral photography [31]. The engagement of students in the peripheral rural areas and with the vulnerable groups, where dental services are less accessible could be effective in terms of student engagement and public health research budget and cost-effectiveness. The COVID-19 pandemic has revealed the importance of studying and refining the method of intraoral photography, to improve the availability of dental services, especially in crucial periods such as that during a pandemic [34, 35]. Based on the existing results, the high sensitivity and specificity of intraoral mobile photography provide a basis for using and integrating artificial intelligence in future studies, where the application will be able to analyze photographs, calculate indices, and develop appropriate recommendations for populations.

In the framework of this study, reliable results were achieved by testing the method of studying S-OHI and MGI, which was rarely reported in the other studies. Thus, this study is important for its diverse results. Considering the psychological attitude of the child to oral examination, the method of intraoral photography is interesting, safe, and reduces the factor of fear and stress in children [36, 37].

A limitation of the study is that we cannot generalize the results of the study to other age groups or residents of rural areas. The potential biases in methodology also need to be considered. A main limitation of using mobile phone photography is the limited protection of patients’ privacy in safeguarding their records. In addition, objective assessment of the e component was not possible with intraoral mobile photography as the researcher examining the photographs did not know the child’s accurate age. Further, a longitudinal design is recommended to study the diagnostic accuracy of intraoral mobile photography by including other age groups and considering the capabilities of newer generation mobile smartphones. Filling such gaps will help make the dental screening process easier, cost effective, and more accessible to different populations.

Conclusions

This study indicates that the use of standardized intraoral mobile photography to assess caries experience intensity, S-OHI, and MGI in children is reliable and holds significant potential as a valuable tool for epidemiological studies in dental public health. Engaging dental students in similar research, guided by future protocols, is advisable, not only to enhance their involvement in scientific research, but also to promote cost-effective and budget-conscious research in dental public health.