Synthetic, non-person related panoramic radiographs created by generative adversarial networks in research, clinical, and teaching applications.

OBJECTIVES: Generative Adversarial Networks (GANs) can produce synthetic images free from personal data. They hold significant value in medical research, where data protection is increasingly regulated. Panoramic radiographs (PRs) are a well-suited modality due to their significant level of standardization while simultaneously displaying a high degree of personally identifiable data. METHODS: We produced synthetic PRs (syPRs) out of real PRs (rePRs) using StyleGAN2-ADA by NVIDIA©. A survey was performed on 54 medical professionals and 33 dentistry students. They assessed 45 radiological images (20 rePRs, 20 syPRs, and 5 syPRcontrols) as real or synthetic and interpreted a single-image syPR according to the image quality (0-10) and 14 different items (agreement/disagreement). They also rated the importance for the profession (0-10). A follow-up was performed for test-retest reliability with >10 % of all participants. RESULTS: Overall, the sensitivity was 78.2 % and the specificity was 82.5 %. For professionals, the sensitivity was 79.9 % and the specificity was 82.3 %. For students, the sensitivity was 75.5 % and the specificity was 82.7 %. In the single syPR-interpretation image quality was rated at a median of 6 and 11 items were considered as agreement. The importance for the profession was rated at a median score of 7. The Test-retest reliability yielded a value of 0.23 (Cohen's kappa). CONCLUSIONS: The study marks a comprehensive testing to demonstrate that GANs can produce synthetic radiological images that even health professionals can sometimes not differentiate from real radiological images, thereby being genuinely considered authentic. This enables their utilization and/or modification free from personally identifiable information. CLINICAL SIGNIFICANCE: Synthetic images can be used for university teaching and patient education without relying on patient-related data. They can also be utilized to upscale existing training datasets to improve the accuracy of AI-based diagnostic systems. The study thereby supports clinical teaching as well as diagnostic and therapeutic decision-making.

Microarchitecture of medication-related osteonecrosis of the jaw (MRONJ); a retrospective micro-CT and morphometric analysis.

Medication-related osteonecrosis of the jaw (MRONJ) is a severe side effect of antiresorptive (AR) drugs such as bisphosphonates (BP) and denosumab (Dmab). Although several risk factors are described, the etiology of MRONJ is still not fully elucidated. Bone-strengthening is the primary aim of antiresorptive therapy; however, overly increased bone mass and microcrack accumulation are also discussed in MRONJ etiologies. The aim of this study is to evaluate the microarchitecture of jaw bones with micro-computed tomography (micro-CT) in AR-treated patients with or without MRONJ. Human jaw bone samples of AR-treated patients were separated into 11 groups by AR treatment bisphosphonate (BP), denosumab (Dmab), both (M) and control groups. Subgroups were divided according to the clinical localization as AR-exposed vital jaw bone (BPexp, Dmabexp, Mexp), osteonecrosis-margin of a sequestrum (BPOmar, DmabOmar, MOmar) and osteonecrosis-sequestrum (BPOseq, DmabOseq, MOseq). Healthy jaw bone (CHB) and osteoporotic jaw bone (COP) represent control groups. Samples underwent retrospective micro-CT and morphometric analysis in representative units by bone volume fraction (BV/TV), bone surface density (BS/BV), trabecular thickness (Tr.Th.), trabecular number (Tr.N.), trabecular space (Tr.Sp.), Euler characteristic for bone connectivity, bone mineral density (BMD) and tissue mineral density (TMD). A total of 141 samples from 78 patients were analyzed. BV/TV of Mexp group (mean: 0.46 ± 0.27) was significantly higher than in the COP group (mean: 0.14 ± 0.05; p = 0.0053). Tr.Th. differed significantly between the BPexp group (mean: 0.32 ± 0.15) and the Mexp group (mean: 0.57 ± 0.20; p = 0.0452) as well as between the BPOseq group (mean: 0.25 ± 0.10) and the MOseq group (mean: 0.39 ± 0.18; p = 0.0417). Signs of trabecular thickening and unorganized trabecular microarchitecture from AR-exposed- to sequestrum groups, were analyzed in 3D reconstructions. However, BS/BV, Tr.N., and Tr.Sp. showed no significant differences. Euler characteristic of the BPOseq group (median: 7.46) doubled compared to that of the BPexp group (median: 14.97; p = 0.0064). Mineralization parameters BMD and TMD were similar in all groups. Findings show evidence of enhanced bone mass and suspect microarchitecture in some AR-treated jaw bone compared to osteoporotic jaw bone. Despite increased bone mass, some MRONJ samples showed decreased trabecular connectivity by Euler characteristic compared to AR-treated jaw bone. These samples may indicate extensive ossification and ineffective bone mass with superficially higher bone mass without existing or even reduced mechanical stability, indicated by connectivity loss. This result might also suggest a high risk to microcrack accumulation. At some point, possibly some kind of over-ossification could lead to under-nourishment and microarchitectural weakness, creating instability, subsequently increasing vulnerability to MRONJ.