Oral Radiology最新文献

Background and purpose: Bitewing radiographs have high diagnostic value in detecting interproximal caries and assessing alveolar bone levels. However, image compression and sensor limitations may compromise image quality and diagnostic value. This study investigates whether two advanced deep-learning methods-a Real-Enhanced Super-Resolution Generative Adversarial Network (Real-ESRGAN) and a Swin Transformer-based image restoration network (SwinIR)-can effectively reconstruct clinically practical details from degraded bitewing images.

Materials and methods: A curated dataset of 4,004 high-quality bitewing radiographs was downsampled to produce paired low- and high-resolution images. Both deep-learning methods were fine-tuned using identical training protocols. Image quality was evaluated using peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM). Additionally, a blinded panel of twelve dentists assessed anatomic clarity, diagnostic accuracy, and perceptual realism using a descriptive rating scale.

Results: The SwinIR achieved superior quantitative results, with a PSNR of 35.56 dB and a SSIM of 0.9287, outperforming Real-ESRGAN, which recorded 31.93 dB and 0.8227, respectively. Clinicians also favored SwinIR for its ability to preserve diagnostic structures such as tooth margins and trabecular patterns, whereas Real-ESRGAN was favored for its perceptual realism and smoother texture rendering.

Conclusion: For bitewing radiographs, the Swin Transformer-based super-resolution model demonstrated more faithful preservation of clinically important structures than Real-ESRGAN, making it the preferred choice when diagnostic precision is prioritized.

Purpose: The primary aim of this study was to evaluate the relationship between maxillary sinus volume (MSV) and morphological variations of the infraorbital canal (IOC) using cone beam computed tomography (CBCT) images. Secondary aims were to determine the frequency of IOC types and accessory canals, and to examine the relationship between them.

Methods: IOC variations were classified according to their protrusion pattern. IOC variations were classified into three patterns: Type 1 (canal embedded in the sinus roof), Type 2 (canal descending below the roof but still attached to it), and Type 3 (canal hanging within the sinus cavity). MSV was calculated using three-dimensional reconstruction. The presence of accessory canals was recorded. A one-way ANOVA was used to evaluate the relationship between IOC variations and MSV, while a chi-square test assessed the association between IOC types and accessory canal presence.

Results: The most frequently observed IOC type was Type 2 (46.5%), followed by Type 1 (29.3%) and Type 3 (24.2%). The mean MSV for Type 1 IOC was lower than for Type 2 and Type 3 (p < 0.001). There was no significant difference in MSV between Type 2 and Type 3 (p > 0.05). No significant relationship was found between the presence of accessory foramina and IOC types (p = 0.612).

Conclusion: Preoperative three-dimensional imaging of MSV and IOC variations is crucial for enhancing surgical precision and minimizing complications in maxillofacial procedures.

Objective: This study aimed to investigate the relationship between the severity of obstructive sleep apnea (OSA) and stiffness of the genioglossus (GG) and geniohyoid (GH) muscles using shear wave elastography (SWE), a non-invasive ultrasonographic technique.

Methods: A total of 102 participants were enrolled, including 51 patients with OSA and 51 healthy controls. Muscle stiffness of the GG and GH was measured using SWE. OSA severity was determined according to the Apnea-Hypopnea Index (AHI) from full-night polysomnography. Comparative and correlation analyses were performed between groups and across severity levels.

Results: GH muscle stiffness was significantly higher in the OSA group than in controls (p = 0.017), whereas GG stiffness showed a non-significant trend (p = 0.061). Among OSA subgroups, GH stiffness was significantly greater in moderate cases than in mild cases, (p = 0.017), but no significant difference was found in GG stiffness across severity levels. Measurement reliability for both muscles was excellent (ICC > 0.95). No significant correlation was observed between stiffness values and daytime sleepiness.

Conclusion: This study demonstrates that GH stiffness increases with OSA and may reflect compensatory muscular adaptations, particularly at moderate disease stages. SWE appears to be a reliable and non-invasive method for assessing upper airway muscle characteristics, offering potential as a supportive tool in OSA evaluation.

Objectives: In the course of a standard diagnostic procedure for temporomandibular joint disorder (TMD), there is often uncertainty regarding the necessity of magnetic resonance imaging (MRI) of the temporomandibular joint (TMJ). This study aims to clarify the relationship between clinical TMD symptoms and MRI findings using logistic regression models, to better define the role of MRI in diagnostics.

Methods: In this retrospective study, the authors analysed a sample of 80 temporomandibular joints (TMJs). Forty patients with TMD symptoms were selected for the study, all of which had previously undergone examination in accordance with the diagnostic criteria for temporomandibular disorders (DC/TMD), as well as having undergone TMJ MRI. Descriptive statistics and regression analyses were used to explore any correlation between clinical symptoms and MRI findings.

Results: MRI-based explanation of clinical symptoms revealed thirteen significant regression models with the following dependent variables: palpation pain at the lateral TMJ pole, TMJ crepitation, condylar hypermobility, uncorrected mandibular deviation, and palpation pain in the medial and lateral pterygoid muscles, as well as in the masseter and temporalis muscles. In contrast, the clinical symptom-based inference of MRI diagnoses yielded eleven significant models, with MRI findings as dependent variables: effusion, degenerative joint disease, anterior disc displacement without reduction, medial disc displacement, thickening at the insertion of the lateral pterygoid muscle, subluxation of the mandibular condyle, reduced glenoid fossa height, and abnormal disc morphology. Among all models, only anterior disc displacement without reduction with condylar hypermobility and with the pain in the masseter muscle demonstrated acceptable predictive accuracy. (AUC = 0.651, AUC = 0.637).

Conclusions: This study confirms that clinical examination alone may be insufficient for accurately diagnosing specific TMJ pathologies. Although some clinical signs show strong associations with MRI findings, only two regression models demonstrated acceptable predictive value.

I am writing to raise concerns regarding the above-mentioned paper, specifically related to the phantom used in the study "Evaluation of the Effect of Reducing Metal Artifacts in Multi-Detector CT Imaging of Zirconia and Titanium Implants" Accepted: February 17, 2025 Oral Radiology https://doi.org/10.1007/s11282-025-00814-5 ". In this paper, the authors state that they created a phantom that I designed and commissioned a specialist company to create. This phantom was designed for CBCT, not CT, and was designed by me to reflect the shape of the mandible to some extent in CT scans, while minimizing artifacts about 15 years ago.　Our original paper was titled "Evaluation of Measurement Accuracy in Diagnostic Imaging Using Cone-Beam CT" and was published in the Journal of the Japanese Society of Oral Implantology (Yoshida Yuri, Morita Yasuhiko, et al., 2009; 22(1):3-14. It is available via open access. Subject: Concern Regarding the Paper "Evaluation of the Effect of Reducing Metal Artifacts in Multi-Detector CT Imaging of Zirconia and Titanium Implants" Accepted: February 17, 2025. I am writing to raise concerns regarding the above-mentioned paper, specifically related to the phantom used in the study. In this paper, the authors state that they created a phantom that I designed and commissioned a specialist company to create. This phantom was designed for CBCT, not CT, and was designed by me to reflect the shape of the mandible to some extent in CBCT Images, while minimizing artifacts.

Background: The accurate and timely diagnosis of dental diseases is critical for effective treatment and improved patient outcomes. However, traditional methods of analyzing panoramic X-ray images rely heavily on the expertise of oral and maxillofacial radiologists and dentists, making the process time-consuming, labor-intensive, and prone to human error. To address these challenges, this study introduces a novel web-based AI application powered by the YOLOv11-TAM model designed to automate the detection and diagnosis of dental diseases from panoramic X-ray images.

Methods: The proposed system integrates a user-friendly interface, a robust PostgreSQL database, and an advanced AI engine based on the YOLOv11-TAM architecture. The AI engine was trained and validated using the publicly available DENTEX dataset, which includes 705 annotated panoramic X-ray images categorized into four disease classes: caries, deep caries, impacted teeth, and periapical lesions. The YOLOv11-TAM model incorporates architectural innovations, including the C3k2 block, Spatial Pyramid Pooling Fast (SPPF) layer, and Transformer-based attention mechanisms, to enhance feature extraction, localization accuracy, and adaptability.

Results: The customized YOLOv11-TAM model demonstrated significant improvements over YOLOv11, achieving about a 15% increase in precision, a high specificity of 0.92, and over 12% improvement in localization accuracy for periapical lesions. Class-specific evaluations revealed superior performance in detecting deep caries and periapical lesions, although challenges remain in diagnosing caries due to class imbalance. The usability study also yielded high satisfaction scores, with an average exceeding 8 across all dimensions, highlighting the application's intuitive design and seamless integration into clinical workflows.

Conclusion: This study presents a transformative web-based AI application that leverages advanced deep learning techniques to enhance the accuracy, efficiency, and accessibility of dental diagnostics. By reducing radiologists' workload and enabling early disease detection, the proposed solution has the potential to revolutionize dental healthcare, particularly in underserved regions.

Objectives: The minimum fascia-tumor distance (MFTD) measured by ultrasound (US) has been proposed for detecting deep lobe parotid tumors, but probe compression may cause measurement bias. Computed tomography (CT), free from such distortion, may provide more reliable measurements. This study compared MFTD values from CT and US in the same cohort.

Methods: We retrospectively analyzed 128 parotid tumor patients who underwent both CT and US before surgery. MFTD values were compared, correlation and agreement were assessed with Spearman coefficients and Bland-Altman plots, and diagnostic accuracy was evaluated using ROC curves.

Results: CT-based MFTD values were significantly larger than US-based values (2.00 ± 2.64 mm vs. 1.49 ± 1.86 mm, P < 0.001), with a mean difference of 0.51 mm. This discrepancy increased with tumor depth. A strong correlation was observed between CT and US (r = 0.936, P < 0.001), but agreement was not optimal. ROC analysis showed comparable performance of CT and US in differentiating deep from superficial lobe tumors (AUC = 0.723 vs. 0.712, P = 0.910), with optimal cutoffs of 4.00 mm for CT and 3.40 mm for US.

Conclusion: CT-based MFTD values were consistently larger than those from US, supporting the presence of probe compression-related bias. Nevertheless, both modalities demonstrated similar accuracy for tumor localization. With appropriate cutoff adjustments, CT and US can both serve as effective tools for preoperative identification of deep lobe parotid tumors.

Level of evidence: 3:

Objective: To assess the effectiveness of deep learning architectures in automatically classifying head positioning errors on panoramic radiographs (PRs).

Methods: A total of 480 anonymized PR images were retrospectively collected and categorized by an experienced oral radiologist based on occlusal plane orientation. The dataset was randomly split into 70% for training, 15% for validation, and 15% for testing to ensure balanced model evaluation. Several pre-trained convolutional neural network (CNN) and vision transformer (ViT) models were fine-tuned using transfer learning (TL) strategy. Models were evaluated using widely adopted performance metrics including accuracy, precision, recall, F1-score, and area under curve based on receiver operating characteristic (ROC-AUC).

Results: Among the tested architectures, ResNet18 achieved the best performance, with an overall test accuracy and F1-score of 0.84. The ViT model demonstrated comparatively lower performance (accuracy and F1-score: 0.77), which may be attributable to the relatively small dataset size. In terms of ROC-AUC performance indicator, ResNet18 also outperformed the ViT model across all classes, revealing superior discriminative capability.

Conclusion: To the best of our knowledge, no prior study has comprehensively evaluated both convolution-based and transformer-based deep learning architectures for detecting occlusal plane positioning errors in PRs. The findings suggest that CNN-based models-particularly ResNet18-can effectively identify such errors. These models may help reduce positioning mistakes, especially in PRs acquired by students or less experienced operators, thereby promoting increased standardization, improved image quality, and enhanced diagnostic reliability in dental practice.

Objective: The aim of this study is to detect the C-shaped canal formation in mandibular second molars on panoramic radiographs based on different Deep Convolutional Neural Networks (DCNNs) trained using panoramic radiographs.

Method: This study includes images of 592 patients, consisting of digital panoramic radiographs and cone-beam computed tomography (CBCT) scans of patients with at least one mandibular second molar, archived in the Department of Oral, Dental, and Maxillofacial Radiology, Faculty of Dentistry, Pamukkale University. To confirm the presence of a C-shaped canal, CBCT images were analyzed and set as the gold standard. From 289 panoramic radiographs with C-shaped canals, a total of 422 mandibular second molars were labeled, and an equal number of 422 mandibular second molars were labeled from 303 panoramic radiographs without C-shaped canals, resulting in a total dataset of 844 labeled panoramic radiographs. To detect C-shaped canals in the 844 panoramic images comprising our dataset, the detection accuracy performance of 11 different deep learning models was investigated. These models were applied to the preprocessed and non-preprocessed panoramic images of mandibular second molars divided into two separate groups as "crown-root" and "root". For the first time in the literature, to the best of our knowledge, model prediction results were fused using majority voting for the detection of C-shaped canals in mandibular second molars. Then, corresponding performance measurements were evaluated in terms of accuracy, precision, recall, specificity and confusion matrices.

Results: For the crown-root dataset, the highest average accuracy metrics for preprocessed and non-preprocessed images were obtained as 0.886 (88.6%) and 0.885 (88.5%), respectively. For the root dataset, the highest average accuracy values for preprocessed and non-preprocessed images were 0.887 (88.7%) and 0.892 (89.2%), respectively. The highest accuracy performance metrics, on the other hand, obtained by the fusion of different DCNNs decisions with the application of majority voting, yielded as 0.902 (90.2%) and 0.897 (89.7%) for crown-root and root dataset groups, respectively.

Conclusion: High-performance values were achieved through the use of combined deep learning architectures. Obtained results show that the proposed method is significant for the detection of C-shaped canals in terms of the success of endodontic treatments, and use of deep learning models are sufficiently capable of assisting clinicians.