Dento maxillo facial radiology最新文献

Objectives: To develop and evaluate JADE, a proof-of-concept retrieval-augmented generation (RAG) diagnostic assistive system, designed to enhance large language model (LLM) reasoning for jawbone lesion assessment. This study examined whether RAG improves diagnostic accuracy and stability compared with standalone LLMs and ORAD, a supervised learning-based system.

Methods: JADE was developed as a cloud-based RAG system integrating an expert-curated oral radiology database embedded using text-embedding-3-large and indexed in Qdrant Cloud. Structured clinical inputs were encoded as prioritized vector queries. Hybrid semantic and keyword-based retrieval appended relevant evidence to prompts for differential diagnosis generation across LLM backbones. Performance was evaluated in 25 validation cases and compared with GPT-5, Claude Sonnet 4.5, DeepSeek-R1, Gemini 2.5 Flash, their RAG configurations, and ORAD. Diagnostic accuracy was analysed using Cochran's Q test, with post-hoc McNemar's tests and Bonferroni correction. Intra-model stability and response time were assessed.

Results: RAG-GPT-5 achieved the highest diagnostic accuracy (20/25), followed by RAG-Claude Sonnet 4.5 (18/25), RAG-DeepSeek R1 (17/25), and RAG-Gemini 2.5 Flash (15/25). Standalone models achieved 9-13/25 correct diagnoses, ORAD achieved 17/25. No significant differences were observed among standalone models or RAG-based models and ORAD. A significant improvement was observed for GPT-5 when integrated with RAG (p = 0.002). RAG configurations showed higher intra-model stability, with RAG-GPT-5 achieving mean stability of 0.90 ± 0.11. Mean response times ranged from 3-10 seconds.

Conclusions: JADE improved diagnostic accuracy and stability compared with standalone LLMs, underscoring value of RAG reasoning in jawbone lesion assessment and marking the first RAG application in dentomaxillofacial radiology.

Objectives: This study investigated the long-term risk of cardiovascular events and all-cause mortality among participants with calcified carotid artery atheroma (CCAA) on panoramic radiographs (PRs) in the Periodontitis and its Relation to Coronary Artery Disease (PAROKRANK) study.

Methods: In the multicentre, multidisciplinary PAROKRANK study, 805 patients with a first myocardial infarction (MI) and 805 matched controls without MI were recruited at 17 hospitals in Sweden. At baseline, the participants were examined with panoramic radiography, and in 737 patients and 739 controls, the carotid artery region was assessable. CCAAs were identified in 251 (34%) patients and 205 (28%) controls at baseline. The primary endpoint was defined as the first occurrence of all-cause mortality, non-fatal MI, non-fatal stroke, or hospitalization following heart failure after a mean follow-up of 10 years. The risks of cardiovascular events and death were evaluated using event survival analysis and regression models.

Results: Participants with bilateral CCAAs, regardless of whether they were patients or controls, had significantly higher mortality and morbidity than those without CCAA (P < 0.001). The risk of cardiovascular events was increased in the presence of bilateral CCAAs among both controls (hazard ratio, 1.96 [95% confidence interval : 1.21-3.16], P = 0.006) and patients (1.67 [1.15-2.43], P = 0.007).

Conclusion: Bilateral CCAAs on PRs were an indicator of an increased risk of future cardiovascular events and early death among both controls and patients in the PAROKRANK study. Therefore, dentists can detect CCAA on PR and contribute to identifying individuals in need of medical attention and treatment of cardiovascular disease to prevent morbidity and early death.

Objectives: This study evaluated a validated artificial intelligence (AI) tool for virtual implant planning by comparing its performance with expert-planned and clinically placed implants in single-tooth rehabilitation cases.

Methods: Pre- and postoperative cone-beam computed tomography (CBCT) scans were retrospectively collected from 32 single-tooth implant cases. Actual implants were placed by expert clinicians using a human intelligence (HI)-based approach guided by the preoperative CBCT. The same scans were used to generate both AI- and HI-based virtual implant plans. Pre- and postoperative scans were registered to allow accurate comparison between placed and planned implants. Three plans were evaluated per case: clinically-placed implant (HI-place), HI-planned implant (HI-plan), and AI-planned implant (AI). Evaluations included angular deviation relative to adjacent teeth and prosthetic wax-ups, vertical and horizontal bone thickness, proximity to vital structures, implant dimensions, planning time, and planning repeatability.

Results: The Friedman test indicated no significant differences in implant location among the three approaches (p > 0.05). Angular deviations were similar across methods for both the neighboring tooth (HI-place=7.01±4.7°; HI-plan=7.30±5.6°; AI = 6.86±3.7°) and the wax-up (HI-place=5.22±3.0°; HI-plan=4.55±3.0°; AI = 5.50±4.9°). Implant diameter and length selections also did not differ significantly across HI-place (4.06±0.4 mm; 10.61±1.4 mm), HI-plan (4.06±0.4 mm; 10.45±1.4 mm), and AI (4.05±0.5 mm; 10.39±1.4 mm). AI was significantly faster (p < 0.05), requiring 150 seconds versus 429 seconds for HI-plan, and showed excellent repeatability with zero deviation between repeated plans.

Conclusions: AI-driven virtual implant planning achieved clinically acceptable placement comparable to expert clinicians in single-implant rehabilitation, with superior efficiency and repeatability, supporting its use in routine presurgical workflows.

Advances in knowledge: Integrating AI for implant location and dimensions selection, with accuracy comparable to expert clinicians, can streamline the digital implant workflow, reduce planning time and allow clinicians to focus on verifying the AI-generated plan prior to clinical transfer, thereby increasing the time available for the actual treatment procedure.

Background: Radiographic imaging is essential in dental diagnostics, yet concerns persist regarding its genotoxic impact. While digital advancements have reduced radiation doses, the effects of X-ray exposure on oral mucosal cells remain debated. This study systematically reviews and quantitatively analyses the effects of X-ray exposure on micronucleus (MN) formation in oral epithelial cells, examining potential age-related variations.

Methods: A thorough literature review was performed across PubMed, Web of Science, EBSCO, and Scopus (updated to November 2024), adhering to PRISMA criteria. Eligible studies examined cytogenetic changes in oral epithelial cells after X-ray exposure in healthy individuals. Data extraction covered study design, imaging modality, radiation dose, MN frequency, and statistical methods. Risk of bias was assessed using a modified Effective Public Health Practice Project tool. A random-effects model synthesized MN frequency changes, and Fisher's Z-transformation analysed age correlations.

Results: Eighteen studies met the inclusion criteria, with 16 centred on panoramic imaging and 2 examining a combination of cone-beam computed tomography and lateral cephalometric X-rays. Meta-analysis confirmed a significant increase in MN frequency post-exposure (standardized mean difference = 0.30, 95% CI: 0.07-0.52, P = .01), indicating genotoxic effects. However, age showed a weak correlation with MN formation (r = 0.149, 95% CI: -0.009 to 0.3, P = .065).

Conclusions: X-ray exposure induces measurable genotoxic damage in oral epithelial cells, though age-related effects remain inconclusive. Adhering to the ALARA principle is crucial to minimizing unnecessary radiation. Future studies should employ larger cohorts and refined biomarkers to enhance risk assessment.

Objectives: Reliable cancellous bone segmentation in cone beam CT (CBCT) images is essential for post-orthognathic assessment of condylar resorption. However, challenges such as edge blurring and low contrast in CBCT images make effective segmentation difficult. This study aims to overcome these issues, providing a foundation for accurate bone quantification to enhance surgical planning and patient outcomes.

Methods: We propose a novel approach to enhance edge-based segmentation for cancellous bone in CBCT images. By incorporating edge features from the cancellous bone region and utilizing cancellous edge localization as an auxiliary task via dual-branch fusion network (DBF-Net), our model leverages shared feature parameters across functions to improve segmentation accuracy and robustness.

Results: Our DBF-Net outperformed other models, achieving Dice coefficient of 91.48%. And the 95% Hausdorff distance decreased to 3.88 mm, demonstrating significant improvement in cancellous bone boundary detection, which is crucial for the post-orthognathic assessment of condylar resorption.

Conclusions: This method provides a robust solution for reliable cancellous bone segmentation in CBCT images to support the quantitative assessment of condylar resorption.

Objectives: The objective was to assess the accuracy of different AI tools in providing the right responses for multiple choice questions (MCQs) and the time taken to complete the responses.

Methods: The study included 80 MCQs, each with 4 options and a correct answer related to oral radiology used to assess the knowledge and skill across 5 different domains. The accuracy levels of ChatGPT, ChatGPT-4o (4o), Microsoft Co-pilot, DeepSeek, Gemini, and Meta AI were assessed and compared using the Chi-Square test. In addition, One-way ANOVA was used to compare the response time between different AI chatbots.

Results: Overall, Microsoft Co-pilot and ChatGPT-4o had the highest accuracy. ChatGPT had the fastest response time. Microsoft Co-pilot and DeepSeek, though not significant, had the highest accuracy for knowledge-based and skill-based queries. For accuracy on 5 domains, Microsoft Co-pilot and ChatGPT-4o had an accuracy of 100% for radiographic safety and DeepSeek was more accurate for radiographic diagnosis. Students took more time to respond than the collective time taken by AI chatbots.

Conclusion: Microsoft Co-pilot had an overall higher accuracy, responded more accurately for knowledge-based questions, and was 100% accurate for queries related to radiographic safety. ChatGPT-4o had the second highest accuracy and DeepSeek served better for radiographic diagnosis.

Objectives: This study aimed to assess whether texture feature maps (TFMs) derived from cone beam computed tomography (CBCT) images improve the detection of small simulated condylar defects and to examine the influence of voxel size.

Methods: Eight dry mandibles received 3 spherical defects (0.9, 1.0, 1.2 mm) in each condyle. CBCT scans were performed at voxel sizes of 0.20 mm and 0.40 mm. Sagittal and coronal slices were exported and processed in MaZda software to generate Gray-Level Co-occurrence Matrix (GLCM)-based texture maps (contrast, entropy, etc). Three experienced radiologists independently evaluated randomized images for the presence of cortical defects. Kappa, McNemar, and Fisher tests (α = .05) were used.

Results: Contrast-based TFMs yielded the highest diagnostic performance, with Kappa values ranging from 0.78 to 1.00 and significant improvement over unprocessed images (P < .05). Entropy maps showed the poorest performance. There was no significant effect of voxel size or defect dimension. TFMs enhanced interobserver agreement.

Conclusions: Texture feature maps, especially based on contrast, may enhance visualization of subtle condylar bone defects independent of voxel size. This post-processing approach could provide a low-dose adjunct to improve CBCT diagnostic confidence in temporomandibular joint (TMJ) imaging.

Advances in knowledge: This study introduces the use of texture feature maps in CBCT evaluation of the mandibular condyle, demonstrating that contrast-based mapping enhances detection of subtle bone defects and may support early TMJ diagnosis.

Objectives: This study aimed to conduct a comprehensive evaluation of general-purpose multimodal large language model (LLM) chatbots in oral radiology.

Methods: Ninety text- and image-based oral radiology questions from a Korean dental university were extracted and categorized into six educational contents and two question types. ChatGPT-4o and Gemini 2.0 Flash were evaluated with following items: accuracy with group differences across six contents (using Fisher's exact test with Bonferroni correction, P < .0167), answer consistency across ten repeated outputs (evaluated as the mean agreement and Fleiss' kappa coefficient), and hallucination (evaluated as the mean of the 5-point Global Quality Score assigned by two oral radiologists).

Results: Multimodal LLM chatbots (ChatGPT-4o and Gemini 2.0 Flash) achieved excellent performance on text-based questions with over 80% accuracy but showed limited performance on image-based tasks, with accuracy under 30%. Additionally, image-based tasks exhibited high response variability, and hallucinations were frequently observed, providing incorrect information. These findings suggest that AI chatbots are not yet suitable for reliable use in oral radiology.

Conclusions: This study provided timely insights into the capabilities and limitations of general-purpose multimodal LLM chatbots in the oral radiology, and will serve as a foundation for more safe and effective applications of AI chatbots in the oral radiology field in the future.

Advances in knowledge: This is the first study to comprehensively assess multimodal LLM chatbots in oral radiology. It provides key insights into the performance benchmarks for AI chatbots in oral radiology, promoting the responsible and transparent integration of AI into dental education.

Objectives: Upper airway assessment is crucial due to its impact on craniofacial growth and airway-related disorders. This retrospective observational cross-sectional study aimed to evaluate the upper pharyngeal airway (UPA) space volume and minimum cross-sectional area (MCSA) in a normal sample from Upper Egypt population, analysing age and gender-related variations using cone-beam computed tomography.

Methods: Ninety cone beam computed tomography scans of patients aged 18-60 years were evaluated. The sample was divided into 3 groups (Group I: ≤20 years, Group II: 21-40 years, and Group III: >40 years). Each group was further divided into 2 subgroups according to gender. Nasopharyngeal, oropharyngeal, hypopharyngeal, and total airway volumes, as well as the MCSA, were calculated using Dolphin 3D software. Two-way ANOVA with Bonferroni post hoc test was used for statistical analysis.

Results: The findings revealed significant gender differences. Males exhibited larger airway dimensions across all age groups except for nasopharyngeal volumes in adults over 20 years. A decline in airway volumes and MCSA in individuals over 40 years. The most constricted airway area predominantly occurred in the oropharynx. A strong positive correlation exists between MCSA and airway volumes. Very good inter- and intra-observer agreement.

Conclusions: Upper pharyngeal airway dimensions in a sample of the Upper Egyptian population are influenced by age and gender, highlighting the need for age and gender specific considerations in the diagnosis and treatment.

Advances in knowledge: This study is the first to evaluate age and gender-related variations in UPA volume within the Upper Egypt population. Also, it introduces a discriminant model that provides a probabilistic classification of the site of MCSA with 62.2% accuracy.

Objectives: This review aims to systematically summarize the current studies on the correlation between different degrees of anterior disc displacement (ADD) and temporomandibular joint (TMJ) space.

Materials and methods: Study protocol was carried out based on the Preferred Reporting Items for Systematic Review and Meta-Analysis protocols (PRISMA). The PubMed, Scopus, Web of Science, and China National Knowledge Infrastructure (CNKI) databases were searched to find the relevant articles from database up to February 13, 2025. Studies that evaluated the correlation between ADD and TMJ spaces were included. We excluded intervention studies, case reports, case series, reviews, systematic reviews, opinion articles, book chapters, and studies involving populations with comorbidities potentially affecting the joints. The risk of bias was assessed using the QUADAS-2 tool.

Results: Of the 1109 articles retrieved, a total of 10 articles on the correlation between different degrees of ADD and TMJ spaces were included after evaluation of the title, abstract, and full text. Throughout different degree of ADD, variations in anterior joint space (AJS) and posterior joint space (PJS) exist due to condyle displacement. When the progresses further to ADD without reduction (ADDWoR), the superior joint space (SJS) becomes narrow.

Conclusions: There was a stepwise correlation between the TMJ spaces and the degree of ADD. The AJS and PJS tend to be various in different ADD groups, and the SJS tends to be narrowed in ADDWoR patients. Future studies should standardize methodologies to validate its clinical applicability.