Dental student application of artificial intelligence technology in detecting proximal caries lesions

OBJECTIVES

This study aimed to investigate the caries diagnosis performances of dental students after training with an artificial intelligence (AI) application utilizing deep learning techniques, a type of artificial neural network.

METHODS

A total of 1200 bitewing radiographs were obtained from the institution's database and two specialist dentists labeled the caries lesions in the images. Randomly selected 1000 images were used for training purposes and the remaining 200 radiographs were used to evaluate the caries diagnostic performance of the AI. Then, a convolutional neural network, a deep learning algorithm commonly employed to analyze visual imagery problems, called "You Only Look Once," was modified and trained to detect enamel and dentin caries lesions in the radiographs. Forty dental students were selected voluntarily and randomly divided into two groups. The pre-test results of dental caries diagnosis performances of both groups were recorded. After 1 week, group 2 students were trained using an AI application. Then, the post-test results of both groups were recorded. The labeling duration of the students was also measured and analyzed.

RESULTS

When both groups' pre-test and post-test results were evaluated, a statistically significant improvement was found for all parameters examined except precision score (p < 0.05). However, the trained group's accuracy, sensitivity, specificity, and F1 scores were significantly higher than the non-trained group in terms of post-test scores (p < 0.05). In group 2 (trained group), the post-test labeling time was considerably increased (p < 0.05).

CONCLUSIONS

The students trained by AI showed promising results in detecting caries lesions. The use of AI can also contribute to the clinical education of dental students.

Reproducibility and diagnostic value of a new wedge-guided bitewing image receptor-holding device

OBJECTIVE

To assess the reproducibility of a wedge-guided bitewing image receptor-holding device (IRHD-WG) compared to a commercially available bitewing image receptor-holding device (IRHD-XCP).

METHODS AND MATERIALS

Ten operators randomly acquired bitewing radiographs of four posterior regions (Premolar, Molar 1, Molar 2, Molar 3) distributed in two dry bone skulls and mandibles using 2 IRHDs (IRHD-WG and IRHD-XCP). The radiographs corresponding to the same region and IRHD but acquired by different operators were paired for reproducibility comparison. A total of 360 pairs of radiographs were randomized for analysis. Objective analysis consisted of extracting the pixel values from each radiograph and comparing the values between each pair by intraclass correlation coefficient. For subjective analysis, the IRHDs images were digitally removed. Five oral radiologists blinded for the study conditions classified the radiograph pairs according to their similarity and diagnostic value. The answers' relative frequencies were compared by the chi-square test (α = 0.05).

RESULTS

The IRHD-WG presented a higher reproducibility (Mean = 0.850; SD = 0.144) than the IRHD-XCP (Mean = 0.615; SD = 0.287) in the objective analysis. Radiographs acquired with the IRHD-WG were more similar than those acquired with the IRHD-XCP (p < 0.001). The diagnostic value of bitewing radiographs acquired with the IRHD-WG was considered superior to those acquired with the IRHD-XCP (p < 0.001).

CONCLUSION

For the objective and subjective analyses of reproducibility, the IRHD-WG performed better than the commercial IRHD-XCP.

Classification of Approximal Caries in Bitewing Radiographs Using Convolutional Neural Networks

Dental caries is an extremely common problem in dentistry that affects a significant part of the population. Approximal caries are especially difficult to identify because their position makes clinical analysis difficult. Radiographic evaluation-more specifically, bitewing images-are mostly used in such cases. However, incorrect interpretations may interfere with the diagnostic process. To aid dentists in caries evaluation, computational methods and tools can be used. In this work, we propose a new method that combines image processing techniques and convolutional neural networks to identify approximal dental caries in bitewing radiographic images and classify them according to lesion severity. For this study, we acquired 112 bitewing radiographs. From these exams, we extracted individual tooth images from each exam, applied a data augmentation process, and used the resulting images to train CNN classification models. The tooth images were previously labeled by experts to denote the defined classes. We evaluated classification models based on the Inception and ResNet architectures using three different learning rates: 0.1, 0.01, and 0.001. The training process included 2000 iterations, and the best results were achieved by the Inception model with a 0.001 learning rate, whose accuracy on the test set was 73.3%. The results can be considered promising and suggest that the proposed method could be used to assist dentists in the evaluation of bitewing images, and the definition of lesion severity and appropriate treatments.

Use of smartphones and tablets in radiographic evaluation: accuracy of caries detection on bitewing radiographs

The objective of this study was to compare the accuracy of smartphone and tablet displays for the detection of incipient caries on bitewing radiographs. Forty-six digital bitewing radiographs with a total of 1656 unrestored, readable surfaces were evaluated. The included surfaces extended from the distal surface of each canine to the last posterior contact in each arch. The evaluation excluded nonreadable proximal surfaces. The gold standard for the presence or absence of incipient caries was the consensus of 2 expert oral and maxillofacial radiologists with 15 and 30 years of experience. The radiographs were then independently reevaluated by 2 other professionals with a master's degree in dentistry with an emphasis on oral radiology with more than 5 years of experience (observers 1 and 2). Observers 1 and 2 performed the reevaluations using a notebook computer (Pavilion dm1) with a calibrated LCD monitor as well as 2 smartphones (iPhone 6s and Galaxy Gran2 Duos) and 2 tablet computers (iPad and Galaxy Tab). The presence or absence of incipient dental caries was classified according to the following scale: 1, definitely present; 2, probably present; 3, uncertain; 4, probably absent; and 5, definitely absent. Receiver operating characteristic (ROC) curves were used to determine the accuracy of the diagnoses established by operators 1 and 2. The areas under the ROC curves (A_z) were compared with the binomial exact test at a significance level of 5%. The mean A_z values for the Pavilion dm1 (0.948), iPhone 6s (0.944), Galaxy Gran2 Duos (0.916), iPad (0.949), and Galaxy Tab (0.950) did not present statistically significant differences (P > 0.05). The results suggest that all of the displays used in this study have statistically similar accuracy and can be used to detect incipient caries on bitewing radiographs.

Alveolar bone level at deciduous molars in Flemish children: A retrospective, radiographic study

AIM

The aim of this retrospective radiographic study in Flemish children was to examine the bone level and bone loss around deciduous molars and factors influencing this.

MATERIALS AND METHODS

Two thousand eight hundred ninety six digital intra-oral radiographs of children younger than 18 years old were screened for eligibility. The distance from the cementoenamel junction to the alveolar bone crest was measured, and tooth surfaces were screened for local risk factors that are presumably related to changes in the bone level. A distance >2 mm was defined as bone loss based on previous literature. All measurements were performed by two examiners.

RESULTS

One thousand four hundred ninety one radiographs of 796 patients (mean age 6.46 ± 2.38 years) were included. The distance between the cementoenamel junction and the alveolar bone crest ranged from 0.07 to 2.88 mm, and the mean distance was 0.93 ± 0.37 mm. This distance was positively correlated with age (p < .001). In 3.5% of patients, bone loss was diagnosed. Caries, fillings and pulp pathology were associated with bone loss and higher cementoenamel junction-alveolar bone crest distances (p < .05).

CONCLUSION

This study found a low prevalence of alveolar bone loss in the primary dentition. Both the bone level and bone loss were strongly correlated with local factors.

Visual and radiographic merged-ICDAS caries progression pattern in 2-6 years old Colombian children: Two-year follow-up

AIM

To describe the 2-year clinical and combined with radiographic caries progression pattern in three cohorts of 2-(a), 4-(b), and 6-yr-old children (c).

DESIGN

Baseline sample comprised 600 children (each-cohort: n = 200). Children's dental surfaces were clinically assessed at schools without air-drying (epi) in terms of: Clinical (C) caries lesions (d/D) with ICDASepi-merged system (C-ICDASepi-d/D:Sound/Initial-epi/Moderate/Extensive); fillings (f/F), and missing teeth (m/M). Clinical ICDAS (C) and Radiographic ICDAS (R) indexes were combined (C + R). Caries experience (dmf + DMF) was expressed as: C-dmfs + DMFS (Moderate/Extensive-d/D); C-ICDASepi-dmfs + DMFS (including Initial-epi-d/D); C + R-dmfs + DMFS; and C + R-ICDASepi-dmfs + DMFS. Follow-up caries progression corresponded to more severe caries score.

RESULTS

Two-year follow-up sample was n = 352 (58.7%-baseline sample): a: n = 81; b: n = 131; c: n = 140. Around 100% of children presented C + R-ICDASepi-dmf + DMF ≥ 1, with following C + R-ICDASepi-dmfs + DMFS means: a: 7.3 ± 14.1; b: 12.2 ± 19.5; c: 13.3 ± 16.9. Progression was about 60% in all cohorts, mainly corresponding, in cohort b and c, to occlusal and distal surfaces of lower first/second-molar-primary teeth, and in cohort a to buccal surfaces of upper primary-incisor teeth. The addition of radiographs increased caries mean and prevalence in 10% as compared to only visual examination.

CONCLUSIONS

Children showed a high C + R caries experience progression rate, mainly related to occlusal and proximal surfaces in lower-molar teeth and buccal in upper-incisor teeth.

Characterization and preliminary imaging evaluation of a clinical prototype stationary intraoral tomosynthesis system

PURPOSE

Technological advancements in dental radiography have improved oral care on many fronts, yet diagnostic efficacy for some of the most common oral conditions, such as caries, dental cracks and fractures, and periodontal disease, remains relatively low. Driven by the clinical need for a better diagnostic yield for these and other dental conditions, we initiated the development of a stationary intraoral tomosynthesis (s-IOT) imaging system using carbon nanotube (CNT) x-ray source array technology. Here, we report the system characterization and preliminary imaging evaluation of a clinical prototype s-IOT system approved for human use.

METHODS

The clinical prototype s-IOT system is comprised of a multibeam CNT x-ray source array, high voltage generator, control electronics, collimator cone, and dynamic digital intraoral detector. During a tomosynthesis scan, each x-ray source is operated sequentially at fixed, nominal tube current of 7 mA and user-specified pulse width. Images are acquired by a digital intraoral detector and the reconstruction algorithm generates slice information in real time for operator review. In this study, the s-IOT system was characterized for tube output, dosimetry, and spatial resolution. Manufacturer specifications were validated, such as tube current, kVp, and pulse width. Tube current was measured with an oscilloscope on the analog output of the anode power supply. Pulse width, kVp, and peak skin dose were measured with a dosimeter with ion chamber and high voltage accessory. In-plane spatial resolution was evaluated via measurement of MTF and imaging of a line pair phantom. Spatial resolution in the depth direction was evaluated via artifact spread measurement. The size of the collimated radiation field was evaluated for compliance with FDA regulations. A dental phantom and human specimens of varying pathologies were imaged on a clinical 2D intraoral imaging system as well as s-IOT for comparison and to explore potential clinical applications.

RESULTS

The measured tube current, kVp, and pulse width values were within 3% of the set values. A cumulative peak skin dose of 1.12 mGy was measured for one complete tomosynthesis scan using a 50-ms pulse per projection view. Projection images and reconstruction slices revealed MTF values ranging from 8.1 to 9.3 cycles/mm. Line pair imaging verified this result. The radiation field was found to meet the FDA requirements for intraoral imaging devices. Tomosynthesis reconstruction slice images of the dental phantom and human specimens provided depth resolution, allowing visibility of anatomical features that cannot be seen in the 2D intraoral images.

CONCLUSIONS

The clinical prototype s-IOT device was evaluated and found to meet all manufacturer specifications. Though the system capability is higher, initial investigations are targeting a low-dose range comparable to a single 2D radiograph. Preliminary studies indicated that s-IOT provides increased image quality and feature conspicuity at a dose comparable to a single 2D intraoral radiograph.

In vitro effect of changing the horizontal angulation of X-ray beam on the detection of proximal enamel caries in bitewing radiographs

BACKGROUND

Bitewing radiography is an important modality useful for the evaluation of teeth in patients of various ages and in different stages of tooth eruption. Clinical examination of proximal surfaces for caries may result in false negative results, especially in tight contact areas. Thus, radiography, as an adjunct to clinical examination, is used as a routine diagnostic modality for caries detection.

OBJECTIVES

The objective of this study was to assess the in vitro effect of changing the horizontal angulation of X-ray beam on the detection of proximal enamel caries in bitewing radiographs.

MATERIAL AND METHODS

This in vitro study was conducted on 150 caries-free human premolars (code: p/16/35/9/210). The teeth were randomly divided into 3 groups (n = 50). Group 1 served as the control group and no carious lesions were induced in this group. Teeth in groups 2 and 3 were immersed in demineralizing solution for 2 and 4 months, respectively. After induction of caries and its radiographic confirmation, the teeth were mounted in wax in groups of 3 and bitewing radiographs were obtained at 0°, 5°, 10° and 15° horizontal angles.

RESULTS

The sensitivity values of bitewing radiographs at 0°, 5°, 10° and 15° horizontal angles were 88%, 90%, 88%, 92% in group 2 and 88%, 94%, 94% and 94% in group 3, respectively. The specificity values of bitewing radiographs at 0°, 5°, 10° and 15° horizontal angles were 92%, 86%, 84% and 76%. The accuracy of bitewing radiographs at 0°, 5°, 10° and 15° horizontal angles was 89.3%, 90%, 88.7% and 87.3%, respectively. The highest diagnostic accuracy was obtained at 5° horizontal angle for caries detection; however, the difference in this regard among the tested horizontal angles was not statically significant (p = 0.846).

CONCLUSIONS

Based on the results of this study, changing the horizontal angulation has no significant effect on the detection of proximal enamel caries in bitewing radiographs.

Digital transillumination in caries detection versus radiographic and clinical methods: an in-vivo study

OBJECTIVES

This article aimed to evaluate: (a) the agreement between a near-infrared light transillumination device and clinical and radiographic examinations in caries lesion detection and (b) the reliability of images captured by the transillumination device.

METHODS

Two calibrated examiners evaluated the caries status in premolars and molars on 52 randomly selected subjects by comparing the transillumination device with a clinical examination for the occlusal surfaces and by comparing the transillumination device with a radiographic examination (bitewing radiographs) for the approximal surfaces. Forty-eight trained dental hygienists evaluated and reevaluated 30 randomly selected images 1-month later.

RESULTS

A high concordance between transillumination method and clinical examination (kappa = 0.99) was detected for occlusal caries lesions, while for approximal surfaces, the transillumination device identified a higher number of lesions with respect to bitewing (kappa = 0.91). At the dentinal level, the two methods identified the same number of caries lesions (kappa = 1), whereas more approximal lesions were recorded using the transillumination device in the enamel (kappa = 0.24). The intraexaminer reliability was substantial/almost perfect in 59.4% of the participants.

CONCLUSIONS

The transillumination method showed a high concordance compared with traditional methods (clinical examination and bitewing radiographs). Caries detection reliability using the transillumination device images showed a high intraexaminer agreement. Transillumination showed to be a reliable method and as effective as traditional methods in caries detection.

Ex vivo evaluation of new 2D and 3D dental radiographic technology for detecting caries

OBJECTIVES

Proximal dental caries remains a prevalent disease with only modest detection rates by current diagnostic systems. Many new systems are available without controlled validation of diagnostic efficacy. The objective of this study was to evaluate the diagnostic efficacy of three potentially promising new imaging systems.

METHODS

This study evaluated the caries detection efficacy of Schick 33 (Sirona Dental, Salzburg, Austria) intraoral digital detector images employing an advanced sharpening filter, Planmeca ProMax(®) (Planmeca Inc., Helsinki, Finland) extraoral "panoramic bitewing" images and Sirona Orthophos XG3D (Sirona Dental) CBCT images with advanced artefact reduction. Conventional photostimulable phosphor images served as the control modality. An ex vivo study design using extracted human teeth, ten expert observers and micro-CT ground truth was employed.

RESULTS

Receiver operating characteristic analysis indicated similar diagnostic efficacy of all systems (ANOVA p > 0.05). The sensitivity of the Schick 33 images (0.48) was significantly lower than the other modalities (0.53-0.62). The specificity of the Planmeca images (0.86) was significantly lower than Schick 33 (0.96) and XG3D (0.97). The XG3D showed significantly better cavitation detection sensitivity (0.62) than the other modalities (0.48-0.57).

CONCLUSIONS

The Schick 33 images demonstrated reduced caries sensitivity, whereas the Planmeca panoramic bitewing images demonstrated reduced specificity. XG3D with artefact reduction demonstrated elevated sensitivity and specificity for caries detection, improved depth accuracy and substantially improved cavitation detection. Care must be taken to recognize potential false-positive caries lesions with Planmeca panoramic bitewing images. Use of CBCT for caries detection must be carefully balanced with the presence of metal artefacts, time commitment, financial cost and radiation dose.

Proximal caries lesion detection using the Canary Caries Detection System: an in vitro study

OBJECTIVE

This study investigated the accuracy of the Canary System (CS) to detect proximal caries lesions in vitro, and compared it with conventional methods: International Caries Detection and Assessment System (ICDAS) II and bitewing radiography (BW).

METHODS

Visible proximal surfaces of extracted human teeth were assessed by ICDAS-II before setting them in five manikin mouth models. Then contacting proximal surfaces in mouth models were assessed by BW and CS. Histological validation with polarized-light microscopy served as a gold standard. Pairwise comparisons were performed on area under the curve (AUC), sensitivity, and specificity of the three methods, and corrected using Bonferroni's method. Sensitivities and specificities were compared using a test of proportions and AUC values were compared using DeLong's method.

RESULTS

The CS presented significantly higher sensitivity (0.933) than ICDAS-II (0.733, P = 0.01) and BW (0.267, P < 0.001), and ICDAS-II higher sensitivity than BW (P < 0.001). There were no significant differences between their specificity values: 0.825 (CS), 0.65 (ICDAS-II), and 0.875 (BW). The AUC of CS (0.862) was significantly higher than of ICDAS-II (0.681, P < 0.001) and BW (0.577, P < 0.001).

CONCLUSION

The CS demonstrated greater accuracy in detecting proximal lesions than ICDAS-II and BW, although without significantly higher specificity.

The Effects on Absorbed Dose Distribution in Intraoral X-ray Imaging When Using Tube Voltages of 60 and 70 kV for Bitewing Imaging

Objectives

Efforts are made in radiographic examinations to obtain the best image quality with the lowest possible absorbed dose to the patient. In dental radiography, the absorbed dose to patients is very low, but exposures are relatively frequent. It has been suggested that frequent low-dose exposures can pose a risk for development of future cancer. It has previously been reported that there was no significant difference in the diagnostic accuracy of approximal carious lesions in radiographs obtained using tube voltages of 60 and 70 kV. The aim of this study was, therefore, to evaluate the patient dose resulting from exposures at these tube voltages to obtain intraoral bitewing radiographs.

Material and Methods

The absorbed dose distributions resulting from two bitewing exposures were measured at tube voltages of 60 and 70 kV using Gafchromic^® film and an anatomical head phantom. The dose was measured in the occlusal plane, and ± 50 mm cranially and caudally to evaluate the amount of scattered radiation. The same entrance dose to the phantom was used. The absorbed dose was expressed as the ratio of the maximal doses, the mean doses and the integral doses at tube voltages of 70 and 60 kV.

Results

The patient receives approximately 40 - 50% higher (mean and integral) absorbed dose when a tube voltage of 70 kV is used.

Conclusions

The results of this study clearly indicate that 60 kV should be used for dental intraoral radiographic examinations for approximal caries detection.