Dento maxillo facial radiology最新文献

Objectives: Precise 3D segmentation of paranasal sinuses is essential for accurate diagnosis and treatment. This study aimed to develop a fully automated segmentation algorithm for the paranasal sinuses using the nnU-Net v2 architecture.

Methods: The nnU-Net v2-based segmentation algorithm was developed using Python 3.6.1 and the PyTorch library, and its performance was evaluated on a dataset of 97 cone beam CT (CBCT) scans. Ground truth annotations were manually generated by expert radiologists using the 3D Slicer software, employing a polygonal labelling technique across sagittal, coronal, and axial planes. Model performance was assessed using several quantitative metrics, including accuracy, Dice coefficient (DC), sensitivity, precision, Jaccard index, area under the curve (AUC), and 95% Hausdorff distance (95% HD).

Results: The nnU-Net v2-based algorithm demonstrated high segmentation performance across all paranasal sinuses. DC values were 0.94 for the frontal, 0.95 for the sphenoid, 0.97 for the maxillary, and 0.88 for the ethmoid sinuses. Accuracy scores exceeded 99% for all sinuses. The 95% HD values were 0.51 mm for both the frontal and maxillary sinuses, 0.85 mm for the sphenoid sinus, and 1.17 mm for the ethmoid sinus. Jaccard indices were 0.90, 0.91, 0.94, and 0.80, respectively.

Conclusions: This study highlights the high accuracy and precision of the nnU-Net v2-based CNN model in the fully automated segmentation of paranasal sinuses from CBCT images. The results suggest that the proposed model can significantly contribute to clinical decision-making processes, facilitating diagnostic and therapeutic procedures.

Objective: This study aimed to develop a fully automated and explainable framework for dental age estimation from panoramic radiographs in young individuals.

Methods: A dataset of 1639 radiographs from individuals aged 8 to 23 years was used. The proposed 2-stage pipeline involved: (1) oriented tooth detection using the YOLO11-OBB model and (2) age estimation using deep learning-based regression models with an attention-weighting module to aggregate predictions from individual teeth. Auxiliary features, including the presence of deciduous teeth and sex, were also evaluated for their impact on model performance.

Results: For the first stage, the tooth detection model achieved an F1 score of 0.981, demonstrating accurate tooth localization and identification. In the later stage, the best-performing model, DenseNet-121 with the deciduous teeth feature, achieved a mean absolute error (MAE) of 1.05 ± 0.95 years. Compared to traditional methods, the proposed framework significantly reduced the MAE.

Conclusions: This study developed an explainable, high-performing deep learning framework that offers a promising solution for real-world age estimation in the forensic domain.

Objectives: To compare the number of dissected cervical lymph nodes in the anatomical level III with that in supraomohyoid neck dissection (SOHND) level III affected by the anatomical relationship between the omohyoid muscle and cricoid cartilage using contrast-enhanced CT (CE-CT) images to assess the validity of the current SOHND.

Methods: CE-CT images of the patients who suffered from malignant tumours in the oral and maxillofacial regions were reviewed. The number of cervical lymph nodes both in the anatomical level III (area between the centre of the inferior border of the hyoid bone [HB] and the inferior border of the cricoid cartilage [CC]) and SOHND level III (area between HB and the intersection of the omohyoid muscle and internal jugular vein [OM-IJ]) were recorded, respectively.

Results: The rate of patients whose number of lymph nodes in level III was affected by the positional relationship between the OM-IJ and CC was almost equal in males and females. As for the patients with OM-IJ below the CC, the number of lymph nodes in SOHND level III increased from that of anatomical level III. Females showed significantly higher values than males (P < .05). Meanwhile, for patients with OM-IJ at or above the CC, the number of lymph nodes in SOHND level III decreased from that of anatomical level III.

Conclusions: The number of dissected cervical lymph nodes differed between the SOHND dissection area and levels I, II, and III. In most cases, SOHND dissects more cervical lymph nodes, especially in female patients.

Objectives: To investigate the impact of disinfecting a photostimulable phosphor plate (PSP) receptor with peracetic acid on proximal carious lesions diagnosis. Additionally, to evaluate and measure artefacts caused by disinfection.

Methods: The proximal surfaces of 40 posterior teeth were evaluated by microcomputed tomography to confirm whether they were sound or carious. These teeth were then mounted in 20 phantoms. Using a PSP receptor (Express System), 1 radiograph of each phantom was acquired. Then, 20 disinfections with peracetic acid were performed, corresponding to a disinfection cycle. After the first disinfection cycle, new radiographs were acquired. This process was repeated 9 times, totalling 200 disinfections (10 cycles × 20 disinfections) and 220 radiographs (200 radiographs + 20 baseline radiographs). The radiographs were assessed independently by 5 examiners for carious lesions diagnosis. A radiologist identified and measured the size of the artefacts. Area under the receiver-operating characteristic curve, sensitivity, and specificity were calculated and compared among the number of disinfections by analysis of variance (α = 0.05). The weighted Kappa index evaluated intra- and inter-examiner agreements.

Results: The disinfections did not affect the diagnostic metrics for detecting proximal carious lesions (P > .05). After 200 disinfections, the biggest artefact measured 0.24 cm. Intra- and inter-examiner agreements ranged from moderate to substantial (0.60-0.69) and from fair to moderate (0.34-0.60).

Conclusions: After 200 disinfections, the radiographic detection of proximal carious lesions remains unaffected by the disinfection with peracetic acid, and the artefacts are still small.

Objectives: To conduct a systematic review evaluating the performance of deep learning (DL) tools in osteoporosis screening using dental imaging and to assess whether these models have been implemented in dental practice.

Methods: Search was performed across 7 electronic databases and 2 sources of grey literature. Included studies applied DL algorithms to dental radiographs or CT in adults diagnosed with osteoporosis, using dual-energy X-ray absorptiometry (DXA) or expert assessment as the reference standard. Risk of bias was assessed using the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies 2) tool, alongside an AI-specific checklist and GRADE to evaluate certainty of evidence.

Results: Thirteen studies met the inclusion criteria. DXA was the most common reference standard (n = 08), and panoramic radiography was the predominant imaging modality (n = 12). Accuracy was the most frequently reported metric (n = 12). Most models used pretrained convolutional neural networks, such as VGG16, GoogleNet, ResNet, and AlexNet. ResNet and EfficientNet architectures showed superior performance, particularly when combined with ensemble techniques and clinical covariates. However, no study reported external validation or implementation in dental practice, limiting their applicability.

Conclusions: All DL models demonstrated potential as supportive tools for osteoporosis screening in dentistry. However, the absence of external validation and clinical integration limits their real-world use. Future research should focus on standardization and development of accessible, validated systems.

Advances in knowledge: This systematic review is the first to show that, despite advancements in DL for osteoporosis screening, clinical applicability remains limited. It underscores the need for robust, user-friendly interfaces to facilitate integration into routine dental practice.

Background: The aim of this study is to determine the prevalence, distribution, and follow-up recommendations of incidental findings in cone beam computed tomography (CBCT) scans acquired for endodontic treatment.

Methods: This was a retrospective review of limited field-of-view (FOV) CBCT images for endodontic evaluation obtained at the Stony Brook University School of Dental Medicine between 2018 and 2022. All scans were reviewed by 1 oral and maxillofacial radiology (OMR) resident and all incidental findings were corroborated with the radiology report approved by 1 of 2 board-certified OMRs. Data extracted for each case included age, gender, and reason for scan; and location, diagnosis, and recommended follow-up for each incidental finding. The recommended follow-up consisted of 5 categories: no recommendation, radiographic monitoring, dental specialty evaluation, medical evaluation, or immediate evaluation.

Results: Data was obtained from 565 CBCT scans with 219 males and 346 females. The mean age was 55.4 ± 18.38. There were 349 maxillary and 216 mandibular scans. The most common incidental finding was mild to moderate mucosal disease (21.4%), then periradicular/periapical odontogenic pathology unrelated to the indication for the scan (15.2%). The most common location for an incidental finding was the gnathic bones (62.2%), then the sino-nasal complex (32.5%). The most common follow-up recommendation was dental specialty evaluation (46.2%). Twenty-five findings (2.3%) were subject to immediate follow-up recommendation.

Conclusions: Incidental findings were found in 90.5% of CBCT scans acquired for endodontic purposes. Medical evaluation or immediate evaluation was recommended for a combined 6.4% of incidental findings. Together, these findings highlight the importance of a thorough review of the entire scan by a trained specialist.

Objectives: Positron emission tomography (PET) images are reconstructed using computed tomography (CT)-based attenuation correction, which may cause artefactual errors due to CT artefacts. This study aimed to investigate the effects of metallic dental prostheses and metal artefacts on the incidence of nodular overcorrection in PET/CT images of the oral cavity in patients with head and neck tumours.

Methods: We retrospectively reviewed the F-18 fluorodeoxyglucose PET/CT images of 27 and 20 patients using silicon photomultiplier (SiPM) and photomultiplier tube (PMT) PET, respectively. We evaluated the presence or absence of overcorrection of tumour-like nodular uptake, number of metallic dental prostheses, number of metallic prosthetic blocks, mean CT values, bright-band artefact area/ratio, and dark-band artefact area/ratio.

Results: There was no significant difference in the incidence of overcorrections between SiPM and PMT PET; however, overcorrections in the former were significantly smaller in size than those in the latter. Overcorrection in PMT PET was more likely to occur when the number of metallic dental prostheses, number of metallic dental prosthetic blocks, mean CT value, and bright-band artefact area/ratio were high. However, there was no significant difference in SiPM PET overcorrection between the positive and negative overcorrection groups for the above conditions.

Conclusion: Overcorrection was associated with metallic dental prostheses and CT artefacts; however, the conditions may have scanner-specific characteristics.

Advances in knowledge: Small tumour-like overcorrections appear on SiPM PET, which should be noted because it is more difficult to notice overcorrections. It is important to analyse the images while checking the pre-attenuation correction PET.

Objectives: To evaluate the influence of the battery level of a handheld X-ray device on the objective image quality of radiographs.

Methods: An acrylic-block was radiographed using a SnapShot solid-state sensor and an Eagle handheld X-ray device set at 60 kVp and 0.88 mAs, with the battery at ten different levels (ranging from fully charged to 10%). Six radiographs were acquired at each level. Mean grey values and standard deviations were collected using ImageJ software to assess image brightness, noise, and uniformity. Additionally, an aluminium step-wedge was radiographed under the same conditions for contrast evaluation. Image quality parameters were compared across different battery levels using analysis of variance (ANOVA) with Tukey's post-hoc test (α = 0.05).

Results: Images obtained with a 100% battery level showed higher brightness than those at 50% (P < 0.0001); the battery level did not affect image noise (P = 0.091); uniformity at 20% battery level differed from all other levels (P < 0.0001); and contrast was highest at 90% battery compared to all other levels (P < 0.0001). Overall, the values remained consistent across different battery levels.

Conclusions: Despite some significant differences, the objective image quality remained stable throughout the entire battery charge cycle of the tested handheld X-ray device.

Objectives: The aim of this study was to investigate whether any feature of carotid artery calcification (CAC) detected on panoramic radiographs (PRs) is associated with more severe signs of cardiovascular disease (CVD), as assessed by carotid ultrasound (CUS) including multi-view assessment of carotid intima media thickness (cIMT).

Methods: The present investigation was a retrospective sub-study of the randomized controlled trial visualization of asymptomatic atherosclerotic disease for optimum cardiovascular prevention (VIPVIZA), which included 60-, 50-, and 40-year-old inhabitants of Västerbotten County, each of whom underwent CUS. The present sub-study included 135 participants who had undergone PR for odontological indications within 18 months before or after CUS examination. Findings of CAC on PR were compared with CUS findings of cIMT and carotid plaque. CAC features were categorized into 4 types: single, scattered, vessel width-defining, or vessel-outlining.

Results: Compared to participants without CAC on PR, those with any CAC type on PR exhibited significantly more carotid plaque (80.9% vs 43.2%, P < .001) and a higher average cIMT score (0.83 vs 0.77 mm, P = .013) on CUS. The vessel-outlining CAC group exhibited the most pronounced cIMT and carotid plaque occurrence (P = .011).

Conclusions: All CAC types detected on PR were associated with CVD on CUS, and vessel-outlining CAC indicated more severe CVD. By detecting CAC on PR, especially vessel-outlining CACs, dentists could contribute to the early identification of patients with asymptomatic CVD, and recommend that these patients seek medical attention for preventive treatment.

Advances in knowledge: All types of CAC detected on PR-particularly the vessel-outlining type-are associated with carotid ultrasound findings, including carotid intima media thickness, indicating CVD and increased risk of stroke and myocardial infarction. Thus, dentists can identify patients at increased risk of cardiovascular events by detecting CAC  on PR, with higher diagnostic reliability in cases with vessel-outlining calcification.

Objective: Intraoral ultrasonography (US) is known for its high accuracy in evaluating the depth of invasion (DOI) in tongue squamous cell carcinoma (SCC). However, measurement discrepancies, such as overestimation or underestimation, can occur in certain cases. This study aimed to identify factors affecting the measurement accuracy of intraoral US.

Methods: A total of 96 cases were included in this study. Patients who underwent imaging diagnosis for primary tongue SCC by oral and maxillofacial radiologists, followed by surgical resection by oral surgeons at our institution between April 2014 and March 2022, were included in the study. Cases were classified according to their clinical and histopathological characteristics, including clinical growth pattern, maximum pathological tumour diameter, pathological T classification, and pattern of invasion (Yamamoto-Kohama classification). Measurement accuracy was evaluated using Spearman's rank correlation coefficient and Bland-Altman analysis.

Results: A total of 76 cases were analysed. Intraoral US demonstrated high diagnostic accuracy across various tumour characteristics. However, greater DOI and larger maximum tumour diameter were associated with increased variability in measurements. Exophytic lesions were more likely to be overestimated, whereas lesions with diffuse infiltration tended to be underestimated.

Conclusion: Although intraoral US provides high accuracy in the preoperative evaluation of tongue SCC, clinicians should be aware of specific tumour characteristics that may influence measurement precision, leading to potential overestimation or underestimation.