Detection Of Caries Lesions Research Articles

CONE BEAM COMPUTED TOMOGRAPHY AND ARTIFICIAL INTELLIGENCE. ¿WHERE WE ARE?

In recent years we have seen a huge number of papers presenting the different possibilities for artificial intelligence (AI) in dentistry. Within the next decade, AI will transform the workflow of modern dental practice. Additionally, AI tools can enable healthcare professionals to increase their reliability, effectiveness, and usefulness, nonetheless potential limitations and errors. (1)Systematic reviews concerning artificial intelligence for radiographic imaging detection of caries lesions concluded that AI-based models exhibit good diagnostic performance. AI provides comprehensive, reliable, and accurate image assessment and disease detection, thus facilitating effective and efficient diagnosis and improving the prognosis of dental caries in periapical radiographs. However, the various AI models come with their own set of pros and cons, so the selection of a model should be based on the specific task objectives and requirements. (2, 3) Accordingly, research projects should be meticulously designed to detect the trueperformance of AI models.

DIGITAL RESTORATIVE PROCEDURES IN DENTISTRY

ABSTRACT The integration of digital technologies in restorative dentistry has significantly transformed clinical workflows, enhancing precision, efficiency, and patient outcomes. This article explores the pivotal role of various digital procedures in modern dental practices. Digital imaging techniques have revolutionized diagnostic capabilities, providing high-resolution, detailed visualizations essential for accurate treatment planning. Digital cone beam computed tomography (CBCT) has further refined diagnostic accuracy, enabling three-dimensional assessment of dental structures, which is crucial for implantology and complex restorative cases. Digital caries detection methods offer enhanced early detection of carious lesions, improving preventative care and treatment outcomes. The advent of digital impression systems has streamlined the process of capturing accurate dental impressions, reducing patient discomfort and enhancing the precision of prosthetic restorations. Additionally, digital design-manufacturing (CAD/CAM) systems have facilitated the rapid production of high-quality dental restorations, allowing for same-day procedures and greater customization. Collectively, these digital advancements are reshaping the landscape of restorative dentistry, offering unprecedented opportunities for improving the accuracy, efficiency, and overall success of dental restorations. This review aims to provide an in-depth understanding of the current state of digital restorative procedures in dentistry, with a focus on the latest technological advancements and their clinical implications. Keywords: Dentistry, digital technology, CAD/CAM, CBCT.

Near-infrared light transillumination (NIR-LT) vs clinical-visual inspection/FOTI to detect interproximal caries lesions in vivo.

To evaluate the ability of near-infrared light transillumination (NIR-LT) to detect interproximal enamel and dentinal caries lesions compared to clinical-visual inspection (VI) aided by fibre-optic transillumination (FOTI). From 170 Finnish adolescents aged 15-17 years, 5294 interproximal surfaces of premolars and molars were examined first clinical-visually aided by FOTI (VI&#43;FOTI) using the International Caries Detection and Assessment System (ICDAS) classification. Subsequently, the surfaces were examined using NIR-LT. The extent of lesions was determined using the modified NIR-LT classification based on the Söchtig criteria. For the analyses, data on upper and lower premolars and molars were combined. Distributions of lesions were presented as frequencies. Differences between VI&#43;FOTI and NIR-LT at the tooth and tooth surface levels were analysed by Chi-square and Fisher's exact tests. Sensitivity and specificity of the NIR-LT method to detect any lesion was performed using VI&#43;FOTI as the gold standard. By VI&#43;FOTI, 92.4% surfaces were classified as sound and by NIR-LT, 88.2%. Enamel caries lesions were found on 7.0% of the surfaces by VI&#43;FOTI and on 11.6% by NIR-LT. Enamel lesions identified by NIR-LT were nearly double for all examined teeth groups, except for lower molars it was 1.3-fold. In 66% of the surfaces, the differences between NIR-LT and VI&#43;FOTI findings were statistically significant (p<0.001). The sensitivity for all teeth of NIR-LT was 48.4% and the specificity was 91.1%. Radiation-free NIR-LT method shows considerable potential as a supplementary method for early detection of caries lesions among low caries prevalence adolescents.

Assessment of dental hard tissues and periodontium in children with hemophilia

Relevance. Children suffering from hemophilia primarily receive dental care when they present with acute tooth pain or exacerbation of an existing condition. Due to the rarity (orphan nature) of hereditary coagulopathies, dentists often lack the experience and knowledge to utilize modern methods of prevention and treatment for this patient category. Additionally, the lack of regular dental check-ups for children with hemophilia often leads to a high level of oral morbidity and a need for specific treatments, as indicated by literature data.Materials and methods. The primary methods used to assess oral health in children with hemophilia included detailed questioning, detection of carious lesions in dental hard tissues, assessment of oral hygiene, and the condition of periodontal tissues.Results. Among pediatric patients with hemophilia, poor oral hygiene was associated with a high prevalence of carious lesions (100%) and very high caries intensity values (8.8, ranging from 7 to 11). Additionally, there was a relatively high prevalence of gingival inflammation, accompanied by pronounced gingival bleeding. A direct correlation was identified between the level of oral hygiene and the degree of gingival bleeding in children aged 6-17 years with hemophilia. The deterioration of dental indicators corresponded to the severity of hemophilia, the primary disease in these children.Conclusion. The study demonstrated that pediatric patients with hemophilia, particularly those with moderate to severe forms, had significantly worse indicators of oral hygiene, dental hard tissues, and gingival health compared to generally healthy children. The results indicate the necessity for dental monitoring and rehabilitation for children with hemophilia to detect and effectively treat oral diseases at early stages.

Reliability of impedance spectroscopy versus digital radiograph and ICDAS-II in occlusal caries detection: a prospective clinical trial

The traditional methods in early caries detection had many limitations. So, this study aimed to evaluate the clinical performance of alternating current impedance spectroscopy ACIST in comparison with digital radiograph and ICDAS-II in detection of occlusal carious lesions. Occlusal surfaces of molar and premolar teeth from 40 adult participants were examined by two observers using three diagnostic methods: (1) international caries detection and assessment system (ICDAS-II) (2) digital radiograph (DR) and (3) Cariescan Pro device (ACIST). Agreement analysis and the difference in sensitivities and specificities were evaluated. The results showed an excellent agreement in the different groups. The difference from the visual tactile against ACIST scoring for enamel caries detection, was statistically significant (p = 0.012) and the agreement was moderate (k = 0.509). For dentinal caries the difference was not statistically significant (p > 0.05) and the agreement was similarly moderate (k < 0.6). The difference from the digital radiograph against ACIST scoring, for enamel caries, digital radiography had significantly higher sensitivity and specificity than ACIST (p < 0.001) while for dentinal caries detection and overall, ACIST had higher sensitivity and digital radiography had higher specificity and the difference was statistically significant (p < 0.001). Visual-tactile examination is a considered as feasible and valid technique for occlusal caries detection, digital radiography is superior to ACIST in diagnosing enamel caries, but it could underestimate the caries depth, ACIST is a reliable tool for detecting occlusal caries in dentin.

In vitro accuracy of ultra-low dose cone-beam CT for detection of proximal caries.

This study aimed to assess the accuracy of ultra-low dose (ULD) cone-beam CT (CBCT) for detection of proximal caries. This in vitro study evaluated 104 molar and premolar teeth. The teeth were mounted in dry skulls and underwent CBCT with 4 protocols of high-resolution (HR), normal (NORM), ULD-HR, and ULD-NORM; 78 CBCT images were scored by 3 observers for the presence and penetration depth of caries twice with a 2-week interval using a 5-point Likert scale. The teeth were then sectioned and observed under a stereomicroscope (gold standard). The 4 protocols were compared with each other and with the gold standard. The receiver operating characteristic curve was drawn, and the area under the curve (AUC) was calculated and compared by the Chi-square test (alpha = .05). The interobserver agreement ranged from 0.5233 to 0.6034 for ULD-NORM, 0.5380 to 0.6279 for NORM, 0.5856 to 0.6300 for ULD-HR, and 0.6614 to 0.7707 for HR images. The intra-observer agreement ranged from 0.6027 to 0.8812 for ULD-HR, 0.7083 to 0.7556 for HR, 0.6076 to 0.9452 for ULD-NORM, and 0.7012 to 0.9221 for NORM images. Comparison of AUC revealed no significant difference between NORM and ULD-NORM (P > .05), or HR and ULD-HR (P > .05). The highest AUC belonged to HR (0.8529) and the lowest to NORM (0.7774). Considering the significant reduction in radiation dose in ULD CBCT and its acceptable diagnostic accuracy for detection of proximal caries, this protocol may be used for detection of proximal carious lesions and assessment of their depth.

Developing the Benchmark: Establishing a Gold Standard for the Evaluation of AI Caries Diagnostics.

Background/Objectives: The aim of this study was to establish a histology-based gold standard for the evaluation of artificial intelligence (AI)-based caries detection systems on proximal surfaces in bitewing images. Methods: Extracted human teeth were used to simulate intraoral situations, including caries-free teeth, teeth with artificially created defects and teeth with natural proximal caries. All 153 simulations were radiographed from seven angles, resulting in 1071 in vitro bitewing images. Histological examination of the carious lesion depth was performed twice by an expert. A total of thirty examiners analyzed all the radiographs for caries. Results: We generated in vitro bitewing images to evaluate the performance of AI-based carious lesion detection against a histological gold standard. All examiners achieved a sensitivity of 0.565, a Matthews correlation coefficient (MCC) of 0.578 and an area under the curve (AUC) of 76.1. The histology receiver operating characteristic (ROC) curve significantly outperformed the examiners' ROC curve (p < 0.001). All examiners distinguished induced defects from true caries in 54.6% of cases and correctly classified 99.8% of all teeth. Expert caries classification of the histological images showed a high level of agreement (intraclass correlation coefficient (ICC) = 0.993). Examiner performance varied with caries depth (p ≤ 0.008), except between E2 and E1 lesions (p = 1), while central beam eccentricity, gender, occupation and experience had no significant influence (all p ≥ 0.411). Conclusions: This study successfully established an unbiased dataset to evaluate AI-based caries detection on bitewing surfaces and compare it to human judgement, providing a standardized assessment for fair comparison between AI technologies and helping dental professionals to select reliable diagnostic tools.

Evaluation of near-infrared digital imaging transillumination compared with bitewing radiography for proximal caries detection in children.

Limitations in traditional caries detection tools have driven the development of alternatives methods, focused on the early lesion detection such as near-infrared digital imaging transillumination (NIDIT). The aim of this study was to evaluate the performance of NIDIT compared with bitewing radiography (BWR) in the detection of interproximal carious lesions in children. A retrospective audit of data from children who had NIDIT, BWR and intraoral photographs was conducted. Carious lesions were scored on a tooth surface level with BWR acting as the primary reference for comparison. Accuracy was determined using multi-class area under the curve (AUC), and correlation was determined using Fleiss' Kappa. Data from 499 tooth surfaces involving 44 children were included in this study. The average age across the participants was 86 months (~7 years) with an average dmft (decayed, missing and filled teeth in primary dentition) of 5.29. Multi-class AUC comparing NIDIT to BWR was 0.70. The correlation between NIDIT and BWR was moderate (0.43), whereas the correlation between photographic examination and BWR was 0.30, which is fair. When compared to BWR, NIDIT showed a high specificity but a low sensitivity for proximal caries detection in primary teeth.

Near-infrared imaging in orthodontic intraoral scanners for early interproximal caries detection

Intraoral scanners commonly used in orthodontic offices now offer near-infrared imaging (NIRI) technology, advertised as a screening tool to identify interproximal caries. This study aimed to evaluate the reliability and validity of NIRI detection of interproximal carious lesions in a common intraoral scanner (iTero Element 5D; Align Technology, San Jose, Calif) with and without bitewing radiograph complement, compared with a microcomputed tomography (micro-CT) reference standard. Extracted human posterior teeth (premolars and molars) were selected for early (noncavitated) interproximal carious lesions (n= 39) and sound control surfaces (n= 47). The teeth were scanned via micro-CT for evaluation by 2 blinded evaluators using consensus scoring. The teeth were mounted to simulate anatomic interproximal contacts and underwent a NIRI scan using iTero Element 5D and bitewing radiographs. Two trained, calibrated examiners independently evaluated (1) near-infrared images alone with clinical photograph, (2) bitewing radiograph alone with clinical photograph, and (3) near-infrared images with bitewing radiograph and clinical photograph in combination, after at least a 10-day washout period between each evaluation. Interrater reliability was highest for NIRI alone (k= 0.533) compared with bitewing radiograph alone (k= 0.176) or in combination (k= 0.256). NIRI alone showed high specificity (0.83-0.96) and moderate sensitivity (0.42-0.63) compared with a micro-CT reference standard. Dentin lesions were significantly more reliably detected than enamel lesions. After rigorous training and calibration, NIRI can be used with moderate reliability, high specificity, and moderate sensitivity to detect noncavitated interproximal carious lesions.

Artificial intelligence enabled smart phone app for real-time caries detection on bitewing radiographs.

Diagnosis of proximal caries is a difficult task. Artificial intelligence (AI) enabled diagnosis is gaining momentum. Therefore, it is of interest to evaluate the effectiveness of an artificial intelligence (AI) smart phone application for bitewing radiography towards real-time caries lesion detection. The Efficient Det-Lite1 artificial neural network was used after training 100 radiographic images obtained from the department of Oral Medicine. Trained model was then installed in a Google Pixel 6 (GP6) smartphone as artificial intelligence app. The back-facing mobile phone video camera of GP6 was utilised to detect caries lesions on 100 bitewing radiographs (BWR) with 80 carious lesion in real-time. Two different techniques such as scanning the static BWR on laptop with a moving mobile and scanning the moving radiograph on the laptop with stationery mobile were used. The average value of sensitivity/precision/F1 scores for both the techniques was 0.75/0.846 and 0.795 respectively. AI programme using the rear-facing mobile phone video camera was found to detect 75% of caries lesions in real time on 100 BWR with a precision of 84.6%. Thus, the use of AI with smart phone app is useful for caries diagnosis which is readily accessible, easy to use and fast.

Influence of notebook computers screens and undergraduate level of dental students in the radiographic detection of carious lesions

Aim: To evaluate the influence of notebook computers screens and undergraduate level of dental students in the radiographic detection of carious lesions. Methods: Bitewing digital radiographs were presented to 3rd and 5th year dental students in three different notebooks computers: Notebook 1 with anti-glare screen (1366×768 pixels), Notebook 2 without anti-glare screen (1366×768 pixels), and Notebook 3 with anti-glare screen (1920×1080 pixels). A reference standard based on a consensus analysis was set by three senior professors of Oral Radiology and Cariology. Sensitivity, specificity and accuracy values were measured and submitted to two-way ANOVA at a significance level of 5%. Results: Notebook 2 provided significantly lower sensitivity values (Mean 56.5% ± 2.94) than notebook 3 (71.1% ± 2.82) (p = 0.002). We found no statistically significant differences between the two undergraduate years (p &gt; 0.05). Conclusion: The anti-glare screen of notebook computers screens can influence the radiographic detection of carious lesions, but the undergraduate level of dental students does not influence this diagnostic task.

Artificial intelligence for radiographic imaging detection of caries lesions: a systematic review

BackgroundThe aim of this systematic review is to evaluate the diagnostic performance of Artificial Intelligence (AI) models designed for the detection of caries lesion (CL).Materials and methodsAn electronic literature search was conducted on PubMed, Web of Science, SCOPUS, LILACS and Embase databases for retrospective, prospective and cross-sectional studies published until January 2023, using the following keywords: artificial intelligence (AI), machine learning (ML), deep learning (DL), artificial neural networks (ANN), convolutional neural networks (CNN), deep convolutional neural networks (DCNN), radiology, detection, diagnosis and dental caries (DC). The quality assessment was performed using the guidelines of QUADAS-2.ResultsTwenty articles that met the selection criteria were evaluated. Five studies were performed on periapical radiographs, nine on bitewings, and six on orthopantomography. The number of imaging examinations included ranged from 15 to 2900. Four studies investigated ANN models, fifteen CNN models, and two DCNN models. Twelve were retrospective studies, six cross-sectional and two prospective. The following diagnostic performance was achieved in detecting CL: sensitivity from 0.44 to 0.86, specificity from 0.85 to 0.98, precision from 0.50 to 0.94, PPV (Positive Predictive Value) 0.86, NPV (Negative Predictive Value) 0.95, accuracy from 0.73 to 0.98, area under the curve (AUC) from 0.84 to 0.98, intersection over union of 0.3–0.4 and 0.78, Dice coefficient 0.66 and 0.88, F1-score from 0.64 to 0.92. According to the QUADAS-2 evaluation, most studies exhibited a low risk of bias.ConclusionAI-based models have demonstrated good diagnostic performance, potentially being an important aid in CL detection. Some limitations of these studies are related to the size and heterogeneity of the datasets. Future studies need to rely on comparable, large, and clinically meaningful datasets.ProtocolPROSPERO identifier: CRD42023470708

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

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. 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. 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). 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.

Detection of caries lesions using a water-sensitive STIR sequence in dental MRI

In clinical practice, diagnosis of suspected carious lesions is verified by using conventional dental radiography (DR), including panoramic radiography (OPT), bitewing imaging, and dental X-ray. The aim of this study was to evaluate the use of magnetic resonance imaging (MRI) for caries visualization. Fourteen patients with clinically suspected carious lesions, verified by standardized dental examination including DR and OPT, were imaged with 3D isotropic T2-weighted STIR (short tau inversion recovery) and T1 FFE Black bone sequences. Intensities of dental caries, hard tissue and pulp were measured and calculated as aSNR (apparent signal to noise ratio) and aHTMCNR (apparent hard tissue to muscle contrast to noise ratio) in both sequences. Imaging findings were then correlated to clinical examination results. In STIR as well as in T1 FFE black bone images, aSNR and aHTMCNR was significantly higher in carious lesions than in healthy hard tissue (p < 0.001). Using water-sensitive STIR sequence allowed for detecting significantly lower aSNR and aHTMCNR in carious teeth compared to healthy teeth (p = 0.01). The use of MRI for the detection of caries is a promising imaging technique that may complement clinical exams and traditional imaging.

Potential impact of metal crowns at varying distances from a carious lesion on its detection on cone-beam computed tomography scans with several protocols.

This study evaluated the impact of artifacts generated by metal crowns on the detection of proximal caries lesions in teeth at various distances using cone-beam computed tomography (CBCT). Additionally, the diagnostic impacts of tube current and metal artifact reduction (MAR) were investigated. Thirty teeth were arranged within 10 phantoms, each containing 1 first premolar, 1 second premolar, and 1 second molar. A sound first molar (for the control group) or a tooth with a metal crown was placed. Of the 60 proximal surfaces evaluated, 15 were sound and 45 exhibited enamel caries. CBCT scans were acquired using an OP300 Maxio unit (Instrumentarium, Tuusula, Finland), while varying the tube current (4, 8, or 12.5 mA) and enabling or disabling MAR. Five observers assessed mesial and distal surfaces using a 5-point scale. Multi-way analysis of variance was employed for data comparison, with P<0.05 indicating statistical significance. The area under the curve (AUC) varied from 0.40 to 0.60 (sensitivity: 0.28-0.45, specificity: 0.44-0.80). The diagnostic accuracy was not significantly affected by the presence of a metal crown, milliamperage, or MAR (P>0.05). However, the overall AUC and specificity were significantly lower for surfaces near a crown (P<0.05). CBCT-based caries detection was not influenced by the presence of a metal crown, variations in milliamperage, or MAR activation. However, the diagnostic accuracy was low and was further diminished for surfaces near a crown. Consequently, CBCT is not recommended for the detection of incipient caries lesions.

Impact of using bitewing radiographs alone or in combination with clinical information on treatment decisions.

The combination of different methods has been advocated to increase sensitivity in detecting secondary caries lesions. This cross-sectional study compared the detection of caries lesions around posterior restorations and treatment decisions using bitewing radiographs alone or in combination with clinical information from patient records. The radiographs (n = 212) were randomly distributed into two sequences for assessment across two phases, with a wash-out period of two weeks. In the first phase (X-ray group), the radiographic images were evaluated without clinical information; in the second phase (X-ray/CARS group), the radiographic images were assessed in conjunction with the CARS score (Caries Associated with Restorations or Sealants) and lesion activity. A radiographic classification system for carious lesions around restorations was adapted to classify the bitewing radiographs included in this study. Evaluations were conducted in consensus by a panel of specialists, focusing on the detection of caries around restorations and subsequent treatment decisions. A chi-squared test was used to compare treatment decisions between the groups, and Cohen's kappa coefficient was employed to assess the agreement of scores. The results showed a significant difference in the distribution of decisions regarding the need for restorative intervention between the groups (p < 0.001), with the X-ray/CARS group leading to more decisions favoring restorative intervention. There was a moderate to good agreement of scores (κ = 0.749). The combination of clinical and radiographic information was found to influence the treatment decision-making process by increasing the likelihood of opting for restorative intervention.

Detection of Cavities from Dental Panoramic X-ray Images Using Nested U-Net Models

Dental caries is one of the most prevalent and chronic diseases worldwide. Dental X-ray radiography is considered a standard tool and a valuable resource for radiologists to identify dental diseases and problems that are hard to recognize by visual inspection alone. However, the available dental panoramic image datasets are extremely limited and only include a small number of images. U-Net is one of the deep learning networks that are showing promising performance in medical image segmentation. In this work, different U-Net models are applied to dental panoramic X-ray images to detect caries lesions. The Detection, Numbering, and Segmentation Panoramic Images (DNS) dataset, which includes 1500 panoramic X-ray images obtained from Ivisionlab, is used in this experiment. The major objective of this work is to extend the DNS Panoramic Images dataset by detecting the cavities in the panoramic image and generating the binary ground truth of this image to use as the ground truth for the evaluation of models. These ground truths are revised by experts to ensure their robustness and correctness. Firstly, we expand the Panoramic Images (DNS) dataset by detecting the cavities in the panoramic images and generating the images’ binary ground truth. Secondly, we apply U-Net, U-Net++ and U-Net3+ to the expanded DNS dataset to learn the hierarchical features and to enhance the cavity boundary. The results show that U-Net3+ outperforms the other versions of U-Net with 95% in testing accuracy.

Real-Time Caries Detection of Bitewing Radiographs Using a Mobile Phone and an Artificial Neural Network: A Pilot Study

This pilot study aimed to implement and assess the performance of an experimental artificial intelligence (AI) mobile phone app in the real-time detection of caries lesions on bitewing radiographs (BWRs) with the use of a back-facing mobile phone video camera. The author trained an EfficientDet-Lite1 artificial neural network using 190 radiographic images from the Internet. The trained model was deployed on a Google Pixel 6 mobile phone and used to detect caries on ten additional Internet BWRs. The sensitivity/precision/F1 scores ranged from 0.675/0.692/0.684 to 0.575/0.719/0.639 for the aggregate handheld detection of caries in static BWRs versus the stationary scanning of caries in a moving video of BWRs, respectively. Averaging the aggregate results, the AI app detected—in real time—62.5% of caries lesions on ten BWRs with a precision of 70.6% using the back-facing mobile phone video camera. When combined with the AI app’s relative ease of use and speed and the potential for global accessibility, this proof-of-concept study could quite literally place AI’s vast potential for improving patient care in dentists’ hands.

Enhancing Caries Detection in Bitewing Radiographs Using YOLOv7.

The study aimed to evaluate the impact of image size, area of detection (IoU) thresholds and confidence thresholds on the performance of the YOLO models in the detection of dental caries in bitewing radiographs. A total of 2575 bitewing radiographs were annotated with seven classes according to the ICCMS™ radiographic scoring system. YOLOv3 and YOLOv7 modelswere employed with different configurations, and their performances were evaluated based on precision, recall, F1-score and mean average precision (mAP). Results showed that YOLOv7 with 640 × 640pixel images exhibitedsignificantly superior performance compared to YOLOv3 in terms of precision (0.557 vs. 0.268), F1-score (0.555 vs. 0.375) and mAP (0.562 vs. 0.458), while the recall was significantly lower (0.552 vs. 0.697). The following experiment found that the overall mAPs did not significantly differ between 640 × 640 pixel and 1280 × 1280 pixelimages, for YOLOv7 with an IoU of 50% and a confidence threshold of 0.001 (p = 0.866). The last experiment revealed that the precision significantly increased from 0.570 to 0.593 for YOLOv7 with an IoU of 75% and a confidence threshold of 0.5, but the mean-recall significantly decreased and led to lower mAPs in both IoUs. In conclusion, YOLOv7 outperformed YOLOv3 in caries detection and increasing the image size did not enhance the model's performance. Elevating the IoU from 50% to 75% and confidence threshold from 0.001 to 0.5 led to a reduction of the model's performance, while simultaneously improving precision and reducing recall (minimizing false positives and negatives) for carious lesiondetection in bitewing radiographs.

Performance of near-infrared light transillumination in the detection of occlusal caries lesions in deciduous teeth

BackgroundThis study compared the performance of near-infrared light transillumination (NILT; DIAGNOcam [DC]) in detecting occlusal caries lesions in deciduous molars with the performance of the International Caries Detection and Assessment System (ICDAS), digital radiographic method (RX) and laser fluorescence (DIAGNOdent pen [LFpen]). MethodsFifty-seven extracted deciduous molars with no frank cavitation caries lesions were selected. One site per tooth was evaluated twice each by two examiners using all methods. Then, a histological examination was performed using a stereomicroscope. The performance of the methods was assessed with respect to two thresholds: D1 (all visible caries lesions affecting enamel) and D3 (caries lesions extended into dentin). ResultsA weighted kappa and intraclass correlation coefficients were calculated for intra- and inter-observer reproducibilities, varying from 0.31 (RX) to 0.89 (LFpen). The intra-observer reproducibility at D1 and D3 thresholds varied from 0.51 (ICDAS) to 0.85 (DC) and from 0.30 (ICDAS) to 1.00 (RX), respectively. The inter-observer reproducibility varied from 0.28 (RX) to 0.47 (DC) and from 0.22 (LFpen) to 0.75 (DC), respectively. At the D1 threshold, higher sensitivity and accuracy were found for LFpen (0.76/0.71) and ICDAS (0.65/0.66); and higher specificities were found for RX (0.94), DC (0.79), and ICDAS (0.71). At the D3 threshold, all methods exhibited lower sensitivities (0.13–0.35) and higher specificity, accuracy, and area under the ROC curve values (0.70–1.00). All methods showed a correlation with histological results, especially for ICDAS and DC. ConclusionsNILT technology (DC) and LFpen exhibited similar performance for the detection of occlusal caries lesions in deciduous molars when compared with ICDAS criteria.