Dento maxillo facial radiology最新文献

Objectives: This study explores the application of artificial intelligence (AI), specifically deep learning, in the detection and classification of mandibular fractures using CT scans.

Methods: Data from 459 patients were retrospectively obtained from West China Hospital of Stomatology, Sichuan University, spanning from 2020 to 2023. The CT scans were divided into training, testing, and independent validation sets. This research focuses on training and validating a deep learning model using the nnU-Net segmentation framework for pixel-level accuracy in identifying fracture locations. Additionally, a 3D-ResNet with pre-trained weights was employed to classify fractures into 3 types based on severity. Performance metrics included sensitivity, precision, specificity, and area under the receiver operating characteristic curve (AUC).

Results: The study achieved high diagnostic accuracy in mandibule fracture detection, with sensitivity >0.93, precision >0.79, and specificity >0.80. For mandibular fracture classification, accuracies were all above 0.718, with a mean AUC of 0.86.

Conclusions: Detection and classification of mandibular fractures in CT images can be significantly enhanced using the nnU-Net segmentation framework, aiding in clinical diagnosis.

Objectives: To evaluate deep learning (DL)-based models for detecting periapical bone rarefaction (PBRs) in panoramic radiographs (PRs), analysing their feasibility and performance in dental practice.

Methods: A search was conducted across seven databases and partial grey literature up to November 15, 2024, using Medical Subject Headings and entry terms related to DL, PBRs, and PRs. Studies assessing DL-based models for detecting and classifying PBRs in conventional PRs were included, while those using non-PR imaging or focusing solely on non-PBR lesions were excluded. Two independent reviewers performed screening, data extraction, and quality assessment using the Quality Assessment of Diagnostic Accuracy Studies-2 tool, with conflicts resolved by a third reviewer.

Results: Twelve studies met the inclusion criteria, mostly from Asia (58.3%). The risk of bias was moderate in 10 studies (83.3%) and high in 2 (16.7%). DL models showed moderate to high performance in PBR detection (sensitivity: 26%-100%; specificity: 51%-100%), with U-NET and YOLO being the most used algorithms. Only one study (8.3%) distinguished Periapical Granuloma from Periapical Cysts, revealing a classification gap. Key challenges included limited generalization due to small datasets, anatomical superimpositions in PRs, and variability in reported metrics, compromising models comparison.

Conclusion: This review underscores that DL-based has the potential to become a valuable tool in dental image diagnostics, but it cannot yet be considered a definitive practice. Multicentre collaboration is needed to diversify data and democratize those tools. Standardized performance reporting is critical for fair comparability between different models.

Advances in knowledge: This study represents the first critical synthesis on this theme, examining a group of lesions with complex manifestations that have been neglected in comparable technological development studies, where research focus has usually been limited to radicular cysts. We identified gaps in classification tasks, insufficient use of ethnically diverse and heterogeneous datasets, and the need for multicentric studies. The variability in data reporting prevents transparent comparisons, even precluding our planned meta-analysis. Consequently, we emphasize the necessity for standardized reporting protocols similar to PRISMA for systematic reviews or STARD for diagnostic or prognostic studies, particularly since accuracy metrics remain inadequately documented while critically important.

Digital X-ray sensors have significantly changed dental radiography, enabling faster image acquisition and reducing radiation doses for patients. Despite the advancements in technology, noise in X-ray imaging remains a challenge. In this study, noise was examined using the Noise-Power Spectrum (NPS) and a non-parametric method. Blank images were taken under different exposure times and voltage settings. The analyses show that noise decreases with longer exposure times. Among the examined sensors, 2 showed distinct NPS peaks, and 1 exhibited no relationship between exposure time and noise levels. These results are discussed on terms of specific sensor structures, artefacts and/or unaccessible post-processing algorithms.

Objective: Surgery is the standard treatment for medication-related osteonecrosis of the jaw (MRONJ). However, there are few reports on the appropriate extent of the bone resection. This pilot study explores the feasibility of a new method for estimating the extent of resection using bone single-photon emission CT (SPECT)-standardized uptake value (SUV).

Methods: We retrospectively analysed 8 MRONJ patients who underwent curettage (n = 2), curettage with removal of the separated sequestrum (n = 2), or marginal resection (n = 4) as part of extensive surgery. The resected regions were compared with the regions estimated using SPECT-SUV. The agreement between the SPECT cold region and the resected region was evaluated using the Dice coefficient (defined as the ratio of 2× overlap volume to resected volume plus SPECT cold region volume), overlap ratio, and volume ratio. The inclusion of CT findings (osteolytic, gap- and irregular-type periosteal reactions, and mixed-type osteosclerosis) in the estimated region was also evaluated. Additionally, histopathological findings from 3 marginal resection cases were used to validate the estimated region.

Results: In all cases, the resected region included the cold regions observed on bone SPECT, with radiotracer accumulation confirmed around the resected region. CT-osteolytic regions were included within the estimated region. The Dice coefficient was 0.53 ± 0.10, the overlap ratio was 86.7 ± 7.2%, and the volume ratio was 235.0 ± 74.7%. Histopathological analysis revealed significant osteocyte necrosis in cold regions, whereas areas with an SUV of 9 displayed normal osteocytes, newly formed bone, and mild inflammatory cell infiltration.

Conclusion: This study suggests that the setting of the SPECT cold region using bone SPECT-SUV may allow for the estimation of the extent of resection in early-to-intermediate-stage MRONJ.

Objectives: This study aimed to assess the efficacy of deep learning applications for the detection of nasal bone fracture on X-ray nasal bone lateral view.

Methods: In this retrospective observational study, 2968 X-ray nasal bone lateral views of trauma patients were collected from a radiology centre, and randomly divided into training, validation, and test sets. Preprocessing included noise reduction by using the Gaussian filter and image resizing. Edge detection was performed using the Canny edge detector. Feature extraction was conducted using the gray-level co-occurrence matrix (GLCM), histogram of oriented gradients (HOG), and local binary pattern (LBP) techniques. Several machine learning algorithms namely CNN, VGG16, VGG19, MobileNet, Xception, ResNet50V2, and InceptionV3 were employed for the classification of images into 2 classes of normal and fracture.

Results: The accuracy was the highest for VGG16 and Swin Transformer (79%) followed by ResNet50V2 and InceptionV3 (0.74), Xception (0.72), and MobileNet (0.71). The AUC was the highest for VGG16 (0.86) followed by VGG19 (0.84), MobileNet and Xception (0.83), and Swin Transformer (0.79).

Conclusions: The tested deep learning models were capable of detecting nasal bone fractures on X-ray nasal bone lateral views with high accuracy. VGG16 was the best model with successful results.

Objective: To evaluate the clinical and radiologic features of acute calcific tendinitis of the longus colli (ACTLC).

Methods: This retrospective, cross-sectional study analysed 30 patients diagnosed with ACTLC from January 2013 to December 2022. Two experienced radiologists independently reviewed CT and MR images to confirm the ACTLC diagnosis and document radiologic findings. Clinical data, including symptoms and laboratory results, were also assessed. The study received approval from the institutional ethics committee, with patient consent waived.

Results: The cohort had a mean age of 49 years and included 19 females (63%). All patients presented with acute cervicalgia, and 29 (97%) exhibited calcifications at the C1-C2 level. A novel imaging feature, termed the "beak sign," was observed in 24 of these 29 patients (83%), defined by an acute angle at the margin of calcification pointing towards the C1-C2 intervertebral space. Prevertebral soft-tissue oedema was present in all patients, with 25 (83%) also showing retropharyngeal fluid accumulation. Among the 14 patients who underwent MRI, 11 (79%) exhibited atlantoaxial joint effusion, a feature rarely reported in ACTLC. Follow-up imaging revealed inferior migration of calcifications in 2 patients, with 1 developing a cyst-like appearance in the post-calcific phase-an unreported finding in ACTLC.

Conclusions: This study represents the largest ACTLC cohort confirmed by cross-sectional imaging. Prevertebral calcifications and soft-tissue oedema were consistently observed in all patients, with the majority also exhibiting retropharyngeal fluid accumulation. This article introduces the "beak sign," a novel imaging finding observed in most cases, and identifies atlantoaxial joint effusion as a newly recognized, common imaging manifestation in ACTLC.

Advances in knowledge: This review of 30 patients with acute calcific tendinitis of the longus colli introduces the "beak sign"-an acute angle at the calcification margin pointing towards the C1-C2 intervertebral space-as a novel imaging feature observed in most cases. Additionally, it identifies atlantoaxial joint effusion as a newly recognized, common imaging manifestation of this condition.

Objectives: This study aimed to establish updated diagnostic reference levels (DRLs) for dental imaging modalities in South Korea.

Methods: In cooperation with the Korea Disease Control and Prevention Agency, various types of institutions (dental clinics, dental hospitals, and dental university hospitals) were selected to investigate the status of diagnostic radiation equipment use. Subsequently, over 300 units were randomly selected for each imaging device type (intraoral, panoramic, and cone-beam CT [CBCT]) as measurement samples. DRLs were defined as the 75th percentile of the dose area product distribution. The differences in dose were analysed based on the type of institution, age of use, country of manufacture, and presence of a multifunction device.

Results: The national DRLs for dental imaging established in this survey were as follows: intraoral imaging at 48 mGy·cm2 for adults and 31 mGy·cm2 for children; panoramic imaging at 354 mGy·cm2 for adults and 224 mGy·cm2 for children; and CBCT at 1856 mGy·cm2 for adults and 1350 mGy·cm2 for children. Private dental clinics and hospitals recorded approximately twice the dose levels of university dental hospitals. CBCT devices in dental hospitals and those that have been in used for 5-10 years showed significantly high radiation doses.

Conclusions: The DRLs established through this study were found to be significantly increased, especially in adult and paediatric panoramic radiographs and paediatric CBCT images, compared with those in previous surveys; moreover, they were higher than those in other countries. The findings of this study can serve as a basis for national dose reduction efforts.

Objectives: Temporomandibular disorder (TMD) patients experience a variety of clinical symptoms, and MRI is the most effective tool for diagnosing temporomandibular joint (TMJ) disc displacement. This study aimed to develop a transformer-based deep learning model to generate T2-weighted (T2w) images from proton density-weighted (PDw) images, reducing MRI scan time for TMD patients.

Methods: A dataset of 7226 images from 178 patients who underwent TMJ MRI examinations was used. The proposed model employed a generative adversarial network framework with a TransUNet architecture as the generator for image translation. Additionally, a disc segmentation decoder was integrated to improve image quality in the TMJ disc region. The model performance was evaluated using metrics such as the structural similarity index measure (SSIM), learned perceptual image patch similarity (LPIPS), and Fréchet inception distance (FID). Three experienced oral radiologists also performed a qualitative assessment through the mean opinion score (MOS).

Results: The model demonstrated high performance in generating T2w images from PDw images, achieving average SSIM, LPIPS, and FID values of 82.28%, 2.46, and 23.85, respectively, in the disc region. The model also obtained an average MOS score of 4.58, surpassing other models. Additionally, the model showed robust segmentation capabilities for the TMJ disc.

Conclusion: The proposed model, integrating a transformer and a disc segmentation task, demonstrated strong performance in MR image generation, both quantitatively and qualitatively. This suggests its potential clinical significance in reducing MRI scan times for TMD patients while maintaining high image quality.

Objectives: To investigate the predictability of late cervical lymph node metastasis using radiomics analysis of ultrasonographic images of tongue cancer.

Methods: We selected 128 patients with tongue cancer who underwent intraoral ultrasonography at the pre-treatment, 35 of whom had late cervical lymph node metastasis. Radiomics analysis was used to extract and quantify the image features. Segmentations were performed on three regions: the hypoechoic region with a 3-mm margin (0 + 3-mm margin), the hypoechoic region alone (0-mm margin), and the 3-mm region surrounding the hypoechoic area (3-mm margin). Support vector machine (SVM) and neural network (NNT) were used as the machine learning models, and sensitivity, specificity, and area under the curve (AUC) from the receiver operating characteristic curves were determined for diagnostic performances.

Results: The AUC values in the test group were 0.893, 0.929, and 0.679 for the SVM models with 0 + 3-, 0-, and 3-mm margins, respectively. The AUC values in the test group were 0.905, 0.952, and 0.821 for the NNT models with 0 + 3-, 0-, and 3-mm margins, respectively.

Conclusions: Radiomics analysis and machine learning models using ultrasonographic images of pre-treated tongue cancer with a hypoechoic area (0-mm margin) could be the best models to predict late cervical lymph node metastasis.

Advances in knowledge: This study makes a significant contribution to the tongue cancer treatment because radiomics analysis and machine learning models using ultrasonographic images of before the primary treatment for the tongue cancer could predict late cervical lymph node metastasis with high accuracy.

Objectives: This systematic review aimed to assess the performance of artificial intelligence (AI) in the evaluation of maxillary sinus mucosal alterations in imaging examinations compared to human analysis.

Methods: Studies that presented radiographic images for the diagnosis of paranasal sinus diseases, as well as control groups for AI, were included. Articles that performed tests on animals, presented other conditions, surgical methods, did not present data on the diagnosis of MS or on the outcomes of interest (area under the curve, sensitivity, specificity, and accuracy), compared the outcome only among different AIs were excluded. Searches were conducted in 5 electronic databases and a gray literature. The risk of bias (RB) was assessed using the QUADAS-2 and the certainty of evidence by GRADE.

Results: Six studies were included. The type of study considered was retrospective observational; with serious RB, and a considerable heterogeneity in methodologies. The IA presents similar results to humans, however, imprecision was assessed as serious for the outcomes and the certainty of evidence was classified as very low according to the GRADE approach. Furthermore, a dose-response effect was determined, as specialists demonstrate greater mastery of the diagnosis of MS when compared to resident professionals or general clinicians.

Conclusions: Considering the outcomes, the AI represents a complementary tool for assessing maxillary mucosal alterations, especially considering professionals with less experience. Finally, performance analysis and definition of comparison parameters should be encouraged considering future research perspectives.

Advances in knowledge: AI is a potential complementary tool for assessing maxillary sinus mucosal alterations, however studies are still lacking methodological standardization.