An attempt to generate panoramic radiographs including jaw cysts using StyleGAN3.

OBJECTIVES: The purpose of this study was to generate radiographs including dentigerous cysts by applying the latest generative adversarial network (GAN; StyleGAN3) to panoramic radiography. METHODS: A total of 459 cystic lesions were selected, and 409 images were randomly assigned as training data and 50 images as test data. StyleGAN3 training was performed for 500 000 images. Fifty generated images were objectively evaluated by comparing them with 50 real images according to four metrics: Fréchet inception distance (FID), kernel inception distance (KID), precision and recall, and inception score (IS). A subjective evaluation of the generated images was performed by three specialists who compared them with the real images in a visual Turing test. RESULTS: The results of the metrics were as follows: FID, 199.28; KID, 0.14; precision, 0.0047; recall, 0.00; and IS, 2.48. The overall results of the visual Turing test were 82.3%. No significant difference was found in the human scoring of root resorption. CONCLUSIONS: The images generated by StyleGAN3 were of such high quality that specialists could not distinguish them from the real images.

Two-step deep learning models for detection and identification of the manufacturers and types of dental implants on panoramic radiographs.

The purpose of this study is to develop two-step deep learning models that can automatically detect implant regions on panoramic radiographs and identify several types of implants. A total of 1,574 panoramic radiographs containing 3675 implants were included. The implant manufacturers were Kyocera, Dentsply Sirona, Straumann, and Nobel Biocare. Model A was created to detect oral implants and identify the manufacturers using You Only Look Once (YOLO) v7. After preparing the image patches that cropped the implant regions detected by model A, model B was created to identify the implant types per manufacturer using EfficientNet. Model A achieved very high performance, with recall of 1.000, precision of 0.979, and F1 score of 0.989. It also had accuracy, recall, precision, and F1 score of 0.98 or higher for the classification of the manufacturers. Model B had high classification metrics above 0.92, exception for Nobel's class 2 (Parallel). In this study, two-step deep learning models were built to detect implant regions, identify four manufacturers, and identify implant types per manufacturer.

Can temporomandibular joint osteoarthritis be diagnosed on MRI proton density weighted images with diagnostic support from the latest deep learning classification models?

OBJECTIVES: This study aimed to clarify the performance of magnetic resonance imaging (MRI)-based deep learning classification models in diagnosing temporomandibular joint osteoarthritis (TMJ-OA) and to compare the developed diagnostic assistance with human observers. METHODS: The subjects were 118 patients who underwent MRI for examination of TMJ disorders. One hundred condyles with TMJ-OA and 100 condyles without TMJ-OA were enrolled. Deep learning was performed with four networks (ResNet18, EfficientNet b4, Inception v3, and GoogLeNet) using five-fold cross validation. Receiver operating characteristics (ROC) curves were drawn for each model and diagnostic metrics were determined. The performances of the four network models were compared using Kruskal-Wallis tests and post-hoc Scheffe tests, and ROCs between the best model and human were compared using chi-square tests, with p < 0.05 considered significant. RESULTS: ResNet18 had areas under the curves (AUCs) of 0.91-0.93 and accuracy of 0.85-0.88, which were the highest among the four networks. There were significant differences in AUC and accuracy between ResNet and GoogLeNet (p = 0.0264 and p = 0.0418, respectively). The kappa values of the models were large, 0.95 for ResNet and 0.93 for EfficientNet. The experts achieved similar AUC and accuracy values ​​to the ResNet metrics, 0.94 and 0.85, and 0.84 and 0.84, respectively, but with a lower kappa of 0.67. Those of the dental residents showed lower values. There were significant differences in AUCs between ResNet and residents (p < 0.0001) and between experts and residents (p < 0.0001). CONCLUSIONS: Using a deep learning model, high performance was confirmed for MRI diagnosis of TMJ-OA.

A cycle generative adversarial network for generating synthetic contrast-enhanced computed tomographic images from non-contrast images in the internal jugular lymph node-bearing area.

The objectives of this study were to create a mutual conversion system between contrast-enhanced computed tomography (CECT) and non-CECT images using a cycle generative adversarial network (cycleGAN) for the internal jugular region. Image patches were cropped from CT images in 25 patients who underwent both CECT and non-CECT imaging. Using a cycleGAN, synthetic CECT and non-CECT images were generated from original non-CECT and CECT images, respectively. The peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) were calculated. Visual Turing tests were used to determine whether oral and maxillofacial radiologists could tell the difference between synthetic versus original images, and receiver operating characteristic (ROC) analyses were used to assess the radiologists' performances in discriminating lymph nodes from blood vessels. The PSNR of non-CECT images was higher than that of CECT images, while the SSIM was higher in CECT images. The Visual Turing test showed a higher perceptual quality in CECT images. The area under the ROC curve showed almost perfect performances in synthetic as well as original CECT images. In conclusion, synthetic CECT images created by cycleGAN appeared to have the potential to provide effective information in patients who could not receive contrast enhancement.

Transfer learning in diagnosis of maxillary sinusitis using panoramic radiography and conventional radiography.

OBJECTIVES: To clarify the performance of transfer learning with a small number of Waters' images at institution B in diagnosing maxillary sinusitis, based on a source model trained with a large number of panoramic radiographs at institution A. METHODS: The source model was created by a 200-epoch training process with 800 training and 60 validation datasets of panoramic radiographs at institution A using VGG-16. One hundred and eighty Waters' and 180 panoramic image patches with or without maxillary sinusitis at institution B were enrolled in this study, and were arbitrarily assigned to 120 training, 20 validation, and 40 test datasets, respectively. Transfer learning of 200 epochs was performed using the training and validation datasets of Waters' images based on the source model, and the target model was obtained. The test Waters' images were applied to the source and target models, and the performance of each model was evaluated. Transfer learning with panoramic radiographs and evaluation by two radiologists were undertaken and compared. The evaluation was based on the area of receiver-operating characteristic curves (AUC). RESULTS: When using Waters' images as the test dataset, the AUCs of the source model, target model, and radiologists were 0.780, 0.830, and 0.806, respectively. There were no significant differences between these models and the radiologists, whereas the target model performed better than the source model. For panoramic radiographs, AUCs were 0.863, 0.863, and 0.808, respectively, with no significant differences. CONCLUSIONS: This study performed transfer learning using a small number of Waters' images, based on a source model created solely from panoramic radiographs, resulting in a performance improvement to 0.830 in diagnosing maxillary sinusitis, which was equivalent to that of radiologists. Transfer learning is considered a useful method to improve diagnostic performance.

[Evaluation of Fat Suppression Effect and SNR Using Pixel Shift Method of Fat Suppression Images with CHESS and IDEAL in Head and Neck MR Imaging].

PURPOSE: This study aims to evaluate the fat suppression effect on images of the head and neck region using chemical shift selective (CHESS), and iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL). METHOD: A self-made phantom containing oil around the simulated bone marrow and muscle was scanned. The signal-to-noise ratio (SNR) was calculated using the National Electrical Manufacturers Association (NEMA) subtraction and pixel shift methods. Thereafter, the fat suppression effect and SNR were calculated in clinical images using the pixel shift method. RESULT: In both phantom and clinical images, the fat suppression effect was higher using IDEAL. In addition, the SNR of the NEMA subtraction method and the pixel shift method in phantom images was higher in the simulated bone marrow than in the simulated muscle. The SNR of the vertebral body was higher than that of the tongue in the clinical images using IDEAL, and the same tendency was observed in the phantom image evaluation. However, there was a difference in SNR between the phantom and clinical images. CONCLUSION: In the head and neck region, fat-suppressed images using IDEAL showed the same higher fat-suppressing effect as that in a previous study. The SNR for the phantom and the clinical images was different. The SNR calculated using the pixel shift method for the phantom images with IDEAL and the clinical images showed the same tendency. Although there is a difference between the SNRs of phantom and clinical images calculated by the pixel shift method, it is suggested that the method can be used to compare the SNR between tissues such as the vertebral body and the tongue.

The Combination of Panoramic Imaging and Waters' Projection Contributes to the Diagnosis of Odontogenic Maxillary Sinusitis.

The purpose of this study was to evaluate the usefulness of adding Waters' projection to panoramic imaging compared with panoramic imaging or Waters' projection alone. Maxillary sinusitis in 106 patients (206 sinuses) was retrospectively assessed with panoramic imaging, Waters' projection, and computed tomography imaging by two oral radiologists. The diagnostic performance was assessed with computed tomography imaging as the gold standard. Receiver operating characteristic curves and area under the curve values were obtained. Inter- and intra-observer agreement was quantified using weighted kappa coefficients. Observer A performed the same procedure twice (A1 and A2 for the first and second observations, respectively). The accuracies of observers A1, B, and A2 with combination imaging were 0.699, 0.636, and 0.718, respectively. Their area under the curve values with combination imaging were 0.746, 0.640, and 0.771, respectively. Inter-observer agreement was good for Waters' projection (κ, 0.650), and poor for panoramic imaging (κ, 0281). Intra-observer agreement was good for Waters' projection (κ, 0.752), and moderate for panoramic imaging (κ, 0.597). Panoramic imaging was equivalent to Waters' projection for diagnosing maxillary sinusitis. Combination imaging comprising panoramic imaging and Waters' projection can contribute to the diagnosis of odontogenic maxillary sinusitis because of its high sensitivity.

SAPHO syndrome of the temporomandibular joint associated with trismus: a case report and review of the literature.

Synovitis, acne, pustulosis, hyperostosis, and osteitis (SAPHO) syndrome is a disorder characterized by pustular skin lesions and osteoarticular lesions. Mandibular involvement occurs in approximately 10% of the cases, and is often seen as recurrent mandibular osteitis with bone sclerosis, mainly involving the body of the mandible in the head and neck region. Middle cranial base with temporomandibular joint (TMJ) involvement in SAPHO syndrome can be diagnostically challenging because of its rarity. Herein, we present a case of a 37-year-old man who suffered from trismus and dull pain in the left TMJ region. The initial panoramic image revealed spotty osteolysis around the left condylar head. Computed tomography (CT) images showed an osteosclerotic change in the middle cranial base including the TMJ. Magnetic resonance images showed a cortical bone change in the left TMJ without anterior disk displacement, with spotty low signal intensity in the left condyle bone marrow on T2- weighted images. Our initial diagnosis was osteomyelitis of the middle cranial base including the TMJ region. However, antimicrobial therapy, training for TMJ opening, and a surgical procedure were not effective. A detailed medical interview, careful check for skin lesions, and further imaging examinations including bone scintigraphy and chest CT led to the diagnosis of SAPHO syndrome. The possibility of SAPHO syndrome should be considered in patients suspected of osteomyelitis of the middle cranial base including the TMJ with unknown etiology.

Diagnostic usefulness of panoramic imaging of the pterygopalatine fossa: case of a schwannoma causing pterygopalatine fossa expansion.

The pterygopalatine fossa is a small area between the posterior wall of the maxillary sinus and the anterior surface of the pterygoid process of the sphenoid bone. The pterygopalatine fossa can be seen clearly on panoramic imaging. We present the case of a 57-year-old man who exhibited right pterygopalatine fossa expansion on panoramic imaging. Computed tomography (CT), magnetic resonance imaging (MRI), and panoramic imaging all showed a tumor at the right pterygopalatine fossa in this patient. CT indicated that the tumor replaced right retromaxillary fat and displaced the posterior wall of the maxillary sinus. On MRI, the tumor showed intermediate signal intensity at the paranasal area on T1-weighted images, and variable intermediate and high signal intensities on fat-suppressed T2-weighted images. It was eventually diagnosed as a schwannoma. Thus, panoramic imaging can be used for disease screening at the posterior border of the maxilla. Our conclusion is based on this report of a patient with a schwannoma at the posterior wall of the maxillary sinus, which panoramic imaging revealed to have pterygopalatine fossa expansion.

Atypical MRI and Histopathological Findings in Dermoid Cyst.

On images, a dermoid cyst is often described as resembling a "sack of marbles" or "marbles in a bag". Typically, it comprises an inhomogeneity filled with multiple nodules in a fluid matrix on both computed tomography and magnetic resonance imaging (MRI). How it appears, however, will vary depending on its histological contents, which may cause confusion in arriving at a diagnosis. This report describes a dermoid cyst in the floor of the mouth of a 55 year-old woman that showed an atypical internal appearance on MRI. Most of the lesion showed homogeneous high signal intensity on T1 - and T2-weighted images, suggesting that it was derived from fat. A small area within the mass, however, showed moderate signal intensity almost equal to that of muscle on T1-weighted images and high signal intensity on fat-suppressed T2-weighted images. Given the location of the lesion, a dermoid cyst was one possible diagnosis. A lipoma or lipoma variants were also considered, however, based on signal intensity. Histopathological section of the excised specimen revealed a dermoid cyst with sebaceous glands in its walls and keratin in its cavity. Dermoid cysts show variation in their internal structures and contents. Since MRI can reflect such histological variation, signal intensity requires careful interpretation.

Intracranial meningeal melanocytoma diagnosed using an interdisciplinary approach: a case report and review of the literature.

BACKGROUND: Meningeal melanocytoma is a rare pigmented tumor arising from leptomeningeal melanocytes. Patients with this tumor might initially consult a dentist because a mass lesion in Meckel's cave could manifest as dental pain and malocclusion, thereby mimicking temporomandibular disorder. The diagnostic approach, especially using imaging modalities, would be challenging in such cases unless an interdisciplinary approach is used. CASE PRESENTATION: Here, we report a case of a 39-year-old Japanese man who had a history of pain and numbness on the left side of his face and malocclusion for 3 months before the initial visit. The diagnosis was primary intracranial meningeal melanocytoma arising from Meckel's cave. CONCLUSIONS: The process by which the final diagnosis of meningeal melanocytoma was reached highlights the importance of collaboration between the medical and dental disciplines. This case also demonstrates that meningeal melanocytoma has a specific signal pattern on magnetic resonance imaging, including high signal intensity on T1-weighted images and low signal intensity on T2-weighted images.

The intravoxel incoherent motion MRI of lateral pterygoid muscle: a quantitative analysis in patients with temporomandibular joint disorders.

OBJECTIVES: To quantitatively evaluate diffusion and perfusion status of lateral pterygoid muscle (LPM) in patients with temporomandibular joint disorder (TMD) by intravoxel incoherent motion (IVIM) imaging and to correlate with findings on temporomandibular joints (TMJs) by conventional MRI. METHODS: 42 patients with TMD underwent MRI. To assess IVIM parameters, diffusion-weighted imaging was obtained by spin-echo-based single-shot echoplanar imaging. Regions of interest were created on all diffusion-weighted images of the superior belly of the lateral pterygoid (SLP) and inferior belly of the lateral pterygoid (ILP) at b-values 0-500 s mm-2. Then, IVIM parameters, diffusion (D) and perfusion (f) were calculated using biexponential fittings. The correlation of these values with conventional MRI findings on TMJs was investigated. RESULTS: For SLP, the f parameter in TMJs with anterior disc displacement without reduction was significantly higher than that in normal ones (p = 0.015). It was also significantly higher in TMJs with joint effusion than in those without (p = 0.016). On the other hand, for both SLP and ILP, the D parameter significantly increased in TMJs with osteoarthritis compared with those without (p = 0.015 and p = 0.022, respectively). CONCLUSIONS: Pathological changes of LPM in patients with TMD may be quantitatively evaluated by IVIM parameters.

Utility of fusion volumetric images from computed tomography and magnetic resonance imaging for localizing the mandibular canal.

OBJECTIVES: The purpose of this study was to investigate whether CT/MRI fusion volumetric images can improve the detectability of the mandibular canal (MC) compared with CT alone. METHODS: Images of 31 lesions within or close to the mandible using both multislice CT (MSCT) and MRI were gathered from our imaging archives. All lesions underwent MSCT and three-dimensional volumetric interpolated breath-hold examination (3D-VIBE) MRI. Of the 62 hemimandibles, 13 hemimandibles were excluded because the MC passed through a lesion. The remaining 49 hemimandibles were included in this study. Each hemimandible was divided into 3 areas (premolar, molar and retromolar), and 147 areas were evaluated. First, the visibility of the MC on CT or its neurovascular bundle (NVB) on 3D-VIBE was evaluated. Second, in areas in which both the MC and NVB were visible, the relative locations of the NVB on MRI and the position of the MC on CT were assessed using CT/MRI fusion volumetric images. RESULTS: The MC and NVB were clearly visible in 100 (68%) and 144 (98%) of 147 areas on CT and MRI, respectively. All NVBs and MCs were in identical locations, and the NVB on MRI was the same size or smaller than the MC on CT in 79 and 21 areas, respectively. CONCLUSIONS: 3D-VIBE MRI can accurately depict the NVB. Compared with CT alone, CT/MRI fusion volumetric imaging improves MC detectability.

Diffusion tensor imaging of the inferior alveolar nerve using 3T MRI: a study for quantitative evaluation and fibre tracking.

OBJECTIVES: Diffusion tensor imaging (DTI) can provide structural information and objective values for nerves. The aims of this study were to perform quantitative evaluation and fibre tracking of the normal inferior alveolar nerve (IAN) using DTI on 3.0-T MRI. METHODS: DTI was applied to 92 IANs of 46 healthy volunteers. Circular regions of interest (ROIs) were placed on three different positions at the mandibular foramen, second molar and mental foramen of each nerve on apparent diffusion coefficient (ADC) and fractional anisotropy (FA) maps, and the ADC and FA of each ROI were measured. Differences in the values arising from the nerve positions were evaluated. Furthermore, fibre tracking of the IANs was performed by tractography, and the quality of visualization was evaluated. RESULTS: There were no significant differences in the ADC and FA between the right and left sides regardless of the anteroposterior positions. Regarding differences arising from the anteroposterior measurement positions, the ADC and FA showed no significant differences (p > 0.017), except for the ADCs between the positions at the mandibular foramen and mental foramen in the left side (p = 0.0068). Overall, 70 (76%) of the 92 IANs could be visualized fully or partially by tractography. CONCLUSIONS: The ADC and FA of the IAN were successfully obtained from healthy volunteers using DTI and were confirmed to be symmetrical regardless of the measurement positions. DTI is a feasible technique for the quantitative evaluation and visualization of the IAN.

Application of diffusion kurtosis imaging to odontogenic lesions: Analysis of the cystic component.

PURPOSE: To assess the feasibility of applying diffusion kurtosis imaging (DKI) to common odontogenic lesions and to compare its diagnostic ability versus that of the apparent diffusion coefficient (ADC) for differentiating keratocystic odontogenic tumors (KCOTs) from odontogenic cysts. MATERIALS AND METHODS: Altogether, 35 odontogenic lesions were studied: 24 odontogenic cysts, six KCOTs, and five ameloblastomas. The diffusion coefficient (D) and excessive kurtosis (K) were obtained from diffusion-weighted images at b-values of 0, 500, 1000, and 1500 s/mm2 on 3T magnetic resonance imaging (MRI). The combination of D and K values showing the maximum density of the probable density function was estimated. The ADC was obtained (0 and 1000 s/mm2 ). Values for odontogenic cysts, KCOTs, and ameloblastomas were compared. Multivariate logistic regression modeling was performed to assess the combination of D and K model versus ADC for differentiating KCOTs from odontogenic cysts. RESULTS: The mean D and ADC were significantly higher for ameloblastomas than for odontogenic cysts or KCOTs (P < 0.05). The mean K was significantly lower for ameloblastomas than for odontogenic cysts or KCOTs (P < 0.05). The mean values of all parameters for odontogenic cysts and KCOTs showed no significant differences (P = 0.369 for ADC, 0.133 for D, and 0.874 for K). The accuracy of the combination of D and K model (76.7%) was superior to that of ADC (66.7%). CONCLUSION: Use of DKI may be feasible for common odontogenic lesions. A combination of DKI parameters can be expected to increase the accuracy of its diagnostic ability compared with ADC. J. Magn. Reson. Imaging 2016;44:1565-1571.