This prospective comparative study enrolled patients from routine clinical practice who presented to the emergency department, attended the outpatient clinic of the Department of Oral and Maxillofacial Surgery, or were referred by private practitioners between May and August 2025.

Inclusion criteria were: [1] acute mandibular trauma [2], age ≥ 18 years [3], no prior radiation-based imaging for the current trauma episode, and [4] ability to provide informed consent. Exclusion criteria were: [1] pregnancy or lactation [2], known contraindications to MRI (e.g., pacemakers, certain implants) [3], clinically significant comorbidities affecting image quality (e.g., severe neurocognitive disorders) [4], inability to comply with MRI procedures or study protocol (e.g., language barriers, psychological disorders, dementia), and [5] concurrent participation in other clinical trials that could interfere with the imaging study.

All participants in this study underwent both CT and MRI examinations, performed by trained clinical staff and research personnel from the Departments of Radiology and Cranio-Maxillofacial and Oral Surgery.

Ethical approval for this clinical study was obtained from the Cantonal Ethics Commission of Zurich, Switzerland (2024–02307). Written informed consent was secured from all participants, in accordance with the Declaration of Helsinki and its later amendments.

All enrolled participants underwent CT imaging according to the clinically established standard reference. In addition, each patient received an MRI scan on a 3 Tesla MAGNETOM Vidafit system (Siemens Healthineers, Forchheim, Germany) using a dedicated 15-channel mandibular coil (NORAS MRI Products, Hoechberg, Germany) (Fig. 1). Five MRI protocols were acquired at sub-millimeter isotropic resolution: a research application 3D ultrashort echo time (UTE) protocol, 3D double-echo steady-state (DESS), a research application 3D T1-weighted Gradient Echo (GRE) “Dark Bone” protocol, 3D T1-GRE stack-of-stars volume interpolated breath-hold examination (StarVIBE), and 3D fast spin echo short-tau inversion recovery (STIR). The sequence parameters were as follows: UTE: repetition time, 4.6 ms; echo time, 0.04 ms; flip angle, 5 degrees; bandwidth, 1184 Hz/Px; fat suppression, none; Matrix 384 × 384 × 384; voxel size (acquisition), 0.6 × 0.6 × 0.6 mm3; acquisition time, 3:05 min, DESS: repetition time, 11.2 ms; echo time, 4.21 ms; flip angle, 30 degrees; bandwidth, 355 Hz/Px; fat suppression, water excitation normal; Phase encoding direction, R » L; Matrix read/phase 104 × 104; total acceleration factor, 2; voxel size (acquisition), 0.4 × 0.4 × 0.8 mm3; acquisition time, 6:41 min, Dark Bone: repetition time, 6.0 ms; echo time, 2.46 ms; flip angle, 4 degrees; bandwidth, 550 Hz/Px; fat suppression, standard; Phase encoding direction, R » L; Matrix read/phase 320 × 320; total acceleration factor, 2; voxel size (acquisition), 0.7 × 0.7 × 0.9 mm3; acquisition time, 3:13 min, StarVIBE: repetition time, 4.9 ms; echo time, 2.1 ms; flip angle, 9 degrees; bandwidth, 500 Hz/Px; fast fat saturation, standard; Phase encoding direction, R » L; Matrix read/phase 208 × 208; total acceleration factor, none; voxel size (acquisition), 0.7 × 0.7 × 0.5 mm3; acquisition time, 7:52 min, and STIR: repetition time, 3300 ms; echo time, 113 ms; flip angle, T2 var; bandwidth, 425 Hz/Px; fat saturation, standard; Phase encoding direction, L » R; Matrix read/phase 256 × 256; total acceleration factor, 4; voxel size (acquisition), 0.4 × 0.4 × 0.8 mm3; acquisition time, 9:57 min. Imaging data were initially acquired in either axial or coronal planes and reformatted through multiplanar reconstruction to generate views in other orientations.

For magnetic resonance imaging (MRI) in the acute trauma setting, a 15-channel mandibular coil (NORAS MRI Products, Hoechberg, Germany) (A and B) was used on a 3 Tesla MRI system (MAGNETOM VidaFit, Siemens Healthineers, Forchheim, Germany), providing a field of view of 32 x 16 x 16 cm. This coil incorporates a 14 + 1 receiver array integrated into a specialized, positioning system that can be adjusted to accommodate the individual anatomical variations of each patient (C). This adaptability enhances imaging precision, allowing for high-resolution visualization of the dentomaxillofacial complex and supporting effective MR-based trauma assessment

CT and MRI data were stored in DICOM format and analyzed using the local Picture Archiving and Communication System (PACS) in DeepUnity Diagnost (release v.1.1.1.2, Dedalus HealthCare, Bonn, Germany).

The imaging data assessments were performed by three independent observers from different medical specialties with varying levels of experience: Observer A (S.S.) is a board-certified attending in the Department of Cranio-Maxillofacial and Oral Surgery with 13 years of clinical experience; Observer B (E.B.) is a attending physician, board-certified radiologist and licensed dentist at the Institute of Diagnostic and Interventional Radiology with 10 years of experience in radiology, and Observer C (A.A.H.) is a resident in the Department of Cranio-Maxillofacial and Oral Surgery with 5 years of experience. A calibration meeting was held with the principal investigator and the observers to ensure consistency in image interpretation, aiming to standardize the assessment process under uniform viewing conditions. To minimize bias, all observers were blinded to the evaluations of others and to the MRI protocols, and all assessments were conducted in a randomized sequence.

The assessment of mandibular fractures involved determining their presence and exact anatomical localization, with both modalities employing the following scale for categorization: 0, pathology not identified; 1, pathology identified but inaccurately diagnosed; or 2, pathology identified and accurately diagnosed. Simultaneously, for each modality and imaging protocol, the evaluation time needed to achieve the correct classification, knowing that only mandibular trauma needed to be assessed, was recorded.

For each MRI protocol technical image quality was rated with regard to diagnostic confidence, background noise, resolution, artifacts using this modified five-point visual analog scale [13]: 5, excellent, with no restrictions for clinical decision-making; 4, very good, containing no substantial adverse effect for clinical decision-making; 3, average, borderline for clinical decision-making due to the image quality; 2, poor, substantial adverse effect for clinical decision-making; 1, very poor, not suitable for clinical decision-making.

Trauma-region-specific analysis included the evaluation of visibility of fracture lines, delineation of cortical borders, and bone-to-soft-tissue contrast for each protocol using a modified five-point visual analog scale [10, 14]: 5, excellent; fracture lines and cortical borders are sharply defined, optimal bone-to-soft-tissue contrast; 4, good; fracture lines and cortical borders are sharply defined with minor limitations; good bone-to-soft-tissue contrast; 3, fair; fracture lines and cortical borders are visible but less distinct with moderate limitations, intermediate bone-to-soft-tissue contrast; 2, below average; fracture lines and cortical borders are poorly defined, low bone-to-soft-tissue contrast; 1, poor; fracture lines and cortical borders cannot be defined, extremely low bone-to-soft-tissue contrast.

The maximum distance between the mandibular fracture margins was measured in the axial, coronal, and sagittal planes according to the method described by Feuerriegel et al. [14] and subsequently analyzed through protocol-specific inter-modality comparisons.

Descriptive statistics were calculated, including the median, interquartile ranges (IQR), mean, and standard deviation. Inter-observer reliability was evaluated using Krippendorff’s alpha coefficient, with values interpreted according to established guidelines: a coefficient of 1.0 indicates perfect reliability, 0 represents agreement equivalent to chance, and values below 0 suggest systematic disagreement among observers [16]. Quantitative data were analyzed by calculating absolute and relative differences between CT measurements and each MRI protocol-specific measurement. Additionally, intermodal differences were systematically assessed using the Wilcoxon signed-rank test. All statistical tests were two-sided, applying a significance threshold of α = 0.05. All statistical analyses were performed using IBM SPSS Statistics (version 29.0.2.0, IBM, Chicago, IL, USA).