Purpose: To objectively evaluate the image quality of conventional two-dimensional intraoral radiographs acquired with a stationary intraoral tomosynthesis (s-IOT) system under different tube voltage settings and post-processing adjustments. Materials and Methods: Twenty-four radiographs of an acrylic-block phantom were obtained using a s-IOT system coupled to a size-2 CMOS sensor, with standardized exposure parameters (7 mA and 0.16 seconds of exposure) under two conditions: two tube voltages (60 and 70 kVp) and, with ("current image") and without ("current frame") post-processing image export modes. Brightness, noise, and uniformity were quantitatively assessed using ImageJ (U.S National Institutes of Health, Bethesda, MD, USA) software. Twenty-four radiographs of a dental digital quality assurance phantom were obtained under the same conditions to evaluate spatial and contrast resolution, and percentage of radiographic contrast. All image quality parameters were compared between export modes and tube voltages using two-way repeated-measures Analysis of Variance followed by Sidak post-hoc tests (P<0.05). Results: Current image radiographs had higher brightness and uniformity, lower percentage of radiographic contrast and noise compared with current frame radiographs. Spatial resolution ranged from 9 lp/mm at 60 kVp to 10 lp/mm at 70 kVp, irrespective of post-processing adjustments. Contrast resolution was superior in radiographs without post-processing adjustments, regardless of the kVp. Conclusion: Radiographs acquired with the s-IOT system without post-processing adjustments show lower brightness and higher contrast, but increased noise and reduced uniformity compared with post processed images. Spatial resolution was unaffected by post-processing adjustments but was improved by increasing exposure to 70 kVp. (Imaging Sci Dent 20260080)
Purpose: In this study, we aimed to assess cervical lymphadenopathy associated with periodontitis using diffusion-weighted imaging. Materials and Methods: The MRI data of patients examined at our dentistry hospital between April 2017 and March 2018 were retrospectively analyzed. The primary predictor was disease status (presence of periodontitis), and the primary outcomes were the mean size and ADC values of cervical LNs. Age and sex were included as additional variables. Statistical evaluation comprised the Mann-Whitney U test, Spearman's correlation coefficient, and receiver operating characteristic (ROC) curve analysis. A P-value<0.05 was considered statistically significant. Results: Records from 51 patients (aged 31-77 years) and 194 lymph nodes were reviewed. Short-axis diameter was significantly larger and ADC values were significantly higher in the periodontitis group than in the non-periodontitis group (P<0.001). ROC curve analysis demonstrated good discriminative performance, with AUC values of 0.87-0.93 for short-axis diameter and 0.89-0.92 for ADC. Optimal cutoffs were 4.26 mm (level IB) and 5.50 mm (level IIA) for short-axis diameter, and 0.89 & times; 10(-3) mm(2)/s (level IB) and 0.85 & times; 10(-3) mm(2)/s (level IIA) for ADC. Conclusion: Our findings suggest that LN size and ADC values are associated with periodontitis-related lymphadenopathy and may serve as quantitative indicators of inflammatory changes. However, these findings should be interpreted as exploratory, and further prospective studies with external validation are required before clinical implementation.
Purpose: Artificial intelligence (AI)-based approaches to mandibular age estimation are limited by reliance on manually selected anatomical features. Thus, this study aimed to develop and validate a convolutional neural network (CNN)-based model to estimate age from panoramic radiographs through the analysis of mandibular radiomic attributes. Materials and Methods: This was a cross-sectional study based on digital panoramic radiographs of patients aged 2 to 97 years. To develop the segmentation model, 600 radiographic images were manually annotated and used only for training. A U-Net-based CNN was implemented for the semantic segmentation task, successfully processing 7,832 radiographic images. The proposed age estimation model was built on a CNN architecture comprising three convolutional blocks, each consisting of a Conv2D layer followed by a MaxPooling2D layer, enabling progressive, hierarchical extraction of visual features. Results: The model achieved a mean absolute error (MAE) of 6.91 years, and a root mean square error (RMSE) of 9.41 years. The mean coefficient of determination (R2) was 0.799. In the comparison between chronological and predicted age, most observations were distributed close to the identity line. However, a slight increase in dispersion was observed in the older age groups. Conclusion: The findings suggest that the proposed CNN-based model shows promising results for estimating age from panoramic radiographs. Although a modest reduction in predictive precision was noted among older individuals, overall performance remained acceptable. However, these results should be interpreted with caution, given the limited data sources and the absence of external validation. (Imaging Sci Dent 20260079)
Purpose:This study aimed to evaluate the ability of artificial intelligence (AI) to detect common errors in panoramic radiographs. Materials and Methods:This retrospective study utilized a dataset of 2,888 anonymized panoramic radiographs obtained from multiple dental imaging units. Three annotators classified the images into 3 categories: "uneven magnification," "tongue space," and "normal." Three deep learning architectures-Attention Dental X-ray, ResNet50, and MobileNet-were developed and evaluated. Model performance was assessed using accuracy, precision, sensitivity, area under the curve, and F1 score with 5-fold cross-validation. Statistical analysis was performed using Python version 3.9 (Python Software Foundation, Beaverton, USA) with the Scikit-learn, Matplotlib, Seaborn, and PyTorch libraries for model development and evaluation. McNemar's test was used for pairwise model comparisons, and Cohen's kappa was used to assess intra-rater and inter-rater reliability. Results:The Attention Dental X-ray model demonstrated superior performance among all evaluated models. It achieved the highest performance in classifying tongue space errors, with an average accuracy of 82.7%, whereas the ResNet50 and MobileNet models achieved accuracies of 56.9% and 64.1%, respectively. Conclusion:Artificial intelligence models, particularly those incorporating attention mechanisms, show strong potential as supplementary tools for detecting errors on panoramic radiographs, improving image quality, and reducing radiation exposure. This study extends the application of AI beyond disease detection to quality assurance in dental radiography. Future research should focus on real-time integration for immediate operator feedback and on the development of automated error correction methods.
Purpose:This ex vivo proof-of-concept study aimed to develop deep learning (DL)-based super-resolution (SR) models to enhance simulated cone-beam computed tomography (CBCT) images. Materials and Methods:Micro-computed tomography data from 51 extracted teeth were artificially degraded to simulate CBCT images. Three DL models, super-resolution convolutional neural network (SRCNN), local texture estimator (LTE), and Swin Transformer for image restoration (SwinIR), were compared with bicubic interpolation. Image quality was assessed using peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), learned perceptual image patch similarity (LPIPS), and deep image structure and texture similarity (DISTS). Three dentists evaluated sharpness and noise using a 5-point Likert scale. Eight observers assessed crack visibility in 47 images for LTE and bicubic interpolation using a 5-point Likert scale; scores were binarized using high and low thresholds. Results:All models significantly outperformed bicubic interpolation on objective metrics. SwinIR showed the highest PSNR (30.36 ± 2.66), whereas SRCNN achieved the highest SSIM (0.889 ± 0.073). LTE achieved the best LPIPS (0.253 ± 0.101) and DISTS (0.203 ± 0.049). Subjectively, LTE received the highest sharpness ratings (mean. 3.79 ± 0.47), whereas bicubic interpolation received the highest noise ratings (3.97 ± 1.43). LTE significantly improved crack visibility (odds ratio = 1.326, P = 0.006 for the low-threshold analysis; odds ratio = 1.310, P = 0.010 for the high-threshold analysis), with a higher pooled area under the curve (0.81 vs. 0.76, P = 0.063). Conclusion:DL-based SR models can enhance simulated CBCT images, with LTE demonstrating superior perceptual sharpness and crack visibility.
Purpose:This study aimed to evaluate condylar positional changes in patients with facial asymmetry undergoing orthodontic treatment and orthognathic surgery by comparing the affected side (the side toward which the chin deviates) and the non-affected side (the contralateral side) across different treatment stages. Materials and Methods:A retrospective cone-beam computed tomography (CBCT) analysis was conducted on 25 patients with facial asymmetry (menton deviation >4 mm) treated at Taipei Veterans General Hospital. CBCT scans were obtained at 4 timepoints: pre-treatment (T1), post-orthodontic dental decompensation (T2), post-surgery (T3), and post-treatment (T4). Linear measurements of anterior joint space (AJS), superior joint space (SJS), posterior joint space (PJS), and the condylar angle were assessed. Results:At baseline (T1), no significant differences were observed in linear measurements between the affected and non-affected sides; however, the condylar angle was significantly smaller on the affected side (69.6° vs. 76.5°, P < 0.05). Following surgery (T2-T3), AJS increased significantly on both sides (P < 0.05), whereas SJS and PJS changes varied by side. From T3 to T4, significant differences between the affected and non-affected sides emerged in SJS (P < 0.05), PJS (P < 0.05), and condylar angle (P < 0.05), suggesting differential remodeling during post-surgical orthodontic finishing. Overall, orthodontic decompensation alone (T1-T2) did not significantly influence condylar position. Conclusion:Pre-surgical orthodontic treatment did not significantly affect condylar position, whereas orthognathic surgery induced measurable positional and angular changes. Post-surgical orthodontic finishing revealed divergent remodeling patterns between affected and non-affected condyles. Further studies with larger samples and long-term follow-up are necessary to clarify the clinical implications of these changes.
Purpose: To compare ultrasonography (US) measures and symptoms between temporomandibular disorders (TMD) and non-TMD groups and to quantify the incremental diagnostic value of US beyond a clinical-psychosocial core. Materials and Methods: This cross-sectional study included 57 adults (28 with TMD; 29 without TMD) assessed using the Diagnostic Criteria for Temporomandibular Disorders. US measured the lateral condyle-capsule distance and masseter thickness. The clinical-psychosocial core comprised pain intensity and disability, jaw function, oral behaviors, central sensitization, pain distribution, and pressure pain thresholds. Penalized logistic regression with paired nested cross-validation was used to compare 3 models: A(core), B (US only), and C (combined). Performance was evaluated using the area under the receiver operating characteristic curve (AUC) and Brier score. Incremental value was assessed using paired changes in AUC, net reclassification improvement, integrated discrimination improvement, and decisioncurve analysis. Results: Anthropometric characteristics did not differ between groups. US measures showed no between-group differences, whereas psychosocial and pain-related measures did. Model A demonstrated good discrimination (AUC = 0.827). The US-only model performed poorly, and adding US to the core did not improve performance (AUC = 0.781); the paired difference in AUC was-0.051(95% CI:- 0.115 to 0.004). Reclassification and decision-curve analyses favored the core model. Sensitivity analyses (all pressure pain threshold sites and multiple imputation) yielded consistent results. Conclusion: US did not provide incremental diagnostic value beyond a clinical-psychosocial core for TMD classification and tended to worsen reclassification and clinical utility. Routine ultrasonography for general TMD diagnosis appears unwarranted and may be reserved for targeted structural indications. Larger, validated studies are needed.
Purpose:This study evaluated the nasopharyngeal anatomy, particularly the fossa of Rosenmüller (FoR), on cone-beam computed tomography (CBCT) in a Korean population to establish normative reference data by sex and age. Materials and Methods:In CBCT images, FoR was classified into three types (A-C) for image analysis. Measurements of nasopharyngeal dimensions were performed in Types B and C. Sex- and age-related differences were evaluated using chi-square and independent t-tests, and reliability was assessed using the intraclass correlation coefficient. Results:In total, 492 CBCTs (244 males, 248 females; 20-69 years) were included. Type C was the most frequent morphology and increased with age. Types A and B were more prevalent among males than among females, whereas Type C was predominant among females (57.3%) compared with males (34.4%). Asymmetry was more frequent in males (13.9%) than in females (10.1%). Significant sex differences due to the larger males were found in the distance of the torus levatorius, the distance between the sphenopalatine notch and the right torus levatorius, and the horizontal and vertical dimensions of the FoR. No significant side-to-side differences were observed. Reliability was excellent (ICC=0.97). Conclusion:Type C was the most frequent morphology in both sexes, whereas Types A and B were more frequently observed in males than in females. These differences may indirectly contribute to sex-related disparities in nasopharyngeal carcinoma (NPC) incidence. The normative reference values may aid early detection in dental imaging, and further prospective studies including NPC patients are needed to clarify the role of nasopharyngeal morphology.
Purpose:This study evaluated the performance of the YOLOv8m-seg model in detecting and delineating interproximal caries and cervical burnout on bitewing radiographs and examined whether increasing the number of training epochs improved segmentation accuracy and consistency. Materials and Methods:In total, 1,410 bitewing radiographs were annotated using polygon-based masks by a trained dental clinician. The YOLOv8m-seg model was trained for 50, 100, and 150 epochs on 1,128 images and validated on 282 images using the Ultralytics segmentation framework. Model performance was assessed using precision, recall, and mean average precision at intersection-over-union thresholds of 0.5 and 0.5 to 0.95 (mAP0.5, mAP0.5-0.95) for both bounding box and mask outputs. Additional evaluation was conducted on a non-augmented validation subset. Results:Extended training duration was associated with improved segmentation performance. The highest mask mAP0.5-0.95 value was 0.828 at epoch 150. Both box-based precision and recall increased with longer training, whereas mask-based evaluation more accurately reflected the model's ability to delineate the boundaries of caries and cervical burnout. Performance appeared consistent across both classes in the augmented validation split but was reduced in the non-augmented validation subset. Conclusion:The YOLOv8m-seg model demonstrated high diagnostic accuracy in distinguishing proximal caries from cervical burnout on bitewing radiographs. Its mask-based outputs may assist clinicians in early lesion recognition and support improved diagnostic decision-making. Future studies should evaluate model generalizability across broader populations and diverse clinical environments and should prioritize assessment using non-augmented validation sets and independent test datasets.
Purpose:This systematic review evaluated the diagnostic performance of optical coherence tomography (OCT) for the early detection of oral cancer, particularly emphasizing sensitivity, specificity, and overall diagnostic accuracy. Materials and Methods:A comprehensive literature search was conducted across PubMed, ScienceDirect, and Wiley Online Library covering publications from 2015 to 2025, supplemented by manual hand-searching of relevant references. Studies were selected using the Population, Intervention, Comparison, Outcome, Study Design (PICOS) framework. Eligible studies evaluated the diagnostic accuracy of OCT using histopathology as the reference standard. Risk of bias was assessed using the QUADAS-2 tool, and the review methodology followed PRISMA 2020 and PRISMA-DTA guidelines. The review protocol was registered in PROSPERO (CRD420251112254). Results:Seven studies met the predefined inclusion criteria. OCT demonstrated high diagnostic performance (sensitivity: 81.5%-100%, specificity: 68.8%-100%), with diagnostic accuracy reaching up to 100% in certain settings. Studies that incorporated machine learning approaches, including convolutional neural networks and support vector machines, consistently achieved superior diagnostic performance compared with conventional OCT interpretation alone. Overall methodological quality was generally low, with several studies exhibiting moderate to high risk of bias in specific domains. Conclusion:OCT, particularly when augmented with artificial intelligence, demonstrates high diagnostic accuracy as a non-invasive imaging modality for the early detection of oral cancer. Its capability to identify dysplastic and malignant changes at the microstructural level offers meaningful diagnostic advantages over conventional examination methods. Nevertheless, larger-scale studies employing standardized protocols are required to confirm its clinical utility and support integration into routine oral cancer screening and diagnostic pathways.
Purpose:The present study aimed to develop 2 deep learning (DL) systems incorporating detection functions for the diagnosis of carotid artery calcifications (CACs) on panoramic radiographs and to compare their diagnostic performances using CAC-based, side-based, and patient-based evaluations. Materials and Methods:Panoramic radiographs from 290 patients with CACs and 290 control patients without CACs were used to develop 2 detection models: one designed to detect individual CACs across the entire radiograph (System 1) and another designed to detect CACs within the limited bilateral cervical areas (System 2). CAC-based performance was evaluated using recall, precision, and F1-score. Side-based and patient-based performances were assessed using sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and the area under the receiver operating characteristic curve (AUC). Results:For System 1, CAC-based recall, precision, and F1-score were 0.81, 0.68, and 0.74, respectively. For System 2, the corresponding values were 0.90, 0.67, and 0.77. Side-based sensitivity, specificity, and AUC were 0.87, 0.80, and 0.83 for System 1, and 0.93, 0.84, and 0.89 for System 2. Patient-based sensitivity, specificity, and AUC were 0.93, 0.73, and 0.83 for System 1, and 0.95, 0.70, and 0.83 for System 2. Although a relatively large number of false positives were observed in CAC-based assessments, side-based and patient-based performances showed improvement. Conclusion:Side-based and patient-based performances were sufficient when calculated on the basis of CAC-based evaluations for diagnosing CACs on panoramic radiographs. When conducting studies of this type, performance assessments should include side-based and patient-based evaluations in addition to CAC-based analyses.
Purpose:To evaluate the impact of conventional 3-dimensional (3D) conformal radiotherapy on the mandibular condylar process in patients with head and neck cancer using pixel intensity and fractal dimension analyses of digital panoramic radiographs. Materials and Methods:Pre- and post-radiotherapy radiographs from 32 patients were digitally analyzed by a single experienced researcher. Regions of interest were selected in the right mandibular condylar process, specifically at the most constricted area corresponding to the condylar neck. Pixel intensity and fractal dimension analyses were subsequently performed. Results:Comparison between the evaluation periods demonstrated a statistically significant decrease in both mean pixel intensity and fractal dimension after radiotherapy. The mean pixel intensity was 115 ± 35.4 before radiotherapy and 104 ± 34.0 after radiotherapy. The mean fractal dimension values before and after radiotherapy were 1.30 ± 0.110 and 1.23 ± 0.129, respectively. Conclusion:Conventional 3D conformal radiotherapy appears to lead to deleterious changes in the microarchitecture and bone mass of the mandibular condylar process in patients with head and neck cancer. Further studies are warranted to determine whether patients with radiation-induced low bone quality are more susceptible to complications such as mandibular fractures and temporomandibular joint dysfunction.
Purpose:This study was performed to establish a procedure for simulated low-dose cone-beam computed tomography (CBCT) scans and to investigate whether the resulting images are comparable in diagnostic accuracy to those obtained using a clinical low-dose protocol. Materials and Methods:ImageJ was used to manipulate the sinogram data from CBCT scans acquired at 5 mA to mimic a technical setting of 2 mA by adding noise to the Radon-transformed projection data before image reconstruction. Four observers compared the simulated 2 mA CBCT scans with original clinical 2 mA CBCT scans acquired previously. The CBCT images were analysed using a protocol with a ranking scale, and the observers were required to select only 1 category for each variable. The Wilcoxon signed-rank test was used to assess differences between the 2 CBCT scan types, with a significance level of P<0.05. Intra-observer agreement was evaluated using the Cohen kappa. Results:Pairwise observer comparisons of the simulated and clinical low-dose CBCT scans showed no significant differences in image quality. Intra-observer agreement was acceptable, and in 5 comparisons, the results indicated a high degree of agreement. Conclusion:Simulated low-dose CBCT scans can be generated using ImageJ. No significant differences in image quality were observed between simulated and clinical low-dose CBCT scans when evaluating mandibular third molars. These findings suggest that manipulation of sinogram data is a promising radiation-free approach for simulating low-dose images in optimisation efforts.
Purpose:This study proposes an ex vivo imaging protocol using a dental-dedicated magnetic resonance imaging (ddMRI) system and qualitatively and quantitatively evaluates the effects of phosphate-buffered saline (PBS) washing on image quality. Materials and Methods:Four half-maxillae and 4 half-mandibles from human cadaveric donors were scanned using a ddMRI system with a dental-specific coil. Two pulse sequences ("anatomy" and "inflammation") were applied. Each specimen was imaged at 4 time points: immediately after PBS immersion and after 24, 48, and 72 hours of washing, totaling 64 images. Image quality was evaluated using signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and conspicuity of anatomical structures: cortical bone, medullary bone, root contour, soft tissue, and the sinus floor/mandibular canal. Conspicuity was rated by 3 observers and analyzed using the Cochran Q test (α=0.05). Results:Cortical bone, medullary bone, and the sinus floor were depicted in all images; root contours (91.1%) and soft tissue (85.9%) were visible in most images. PBS washing did not significantly impact conspicuity (P>0.05). Signal intensity was higher in "anatomy" than "inflammation." In the "anatomy" sequence, both SNR and CNR initially declined, then stabilized from 24 hours onward. CNR between medullary bone and soft tissue showed the greatest improvement with extended PBS washing, especially for the "inflammation" sequence. Conclusion:The proposed ddMRI protocol enabled consistent visualization of dentomaxillofacial structures in ex vivo samples. PBS represented a suitable carrier. Although PBS washing did not influence conspicuity, at least 24 hours of washing may improve image quality, as reflected by SNR and CNR.
