Panoramical radiographs or cone-beam computed tomography are used for preoperative planning in orthodontics and for endo/peridontic procedures but do not allow assessment of the mandibular nerve. MRI is suited for assessment of neuronal integrity, but challenging when imaging cortical bone or the mandibular nerve. Cortical bone yields almost no signal with standard MR sequences such as fast spin- or gradient-echo imaging [1]. In contrast, ultra-short echo time image acquisition generates detectable signals from osseous structures [2]. Concerning morphological neurography of thin peripheral nerves, commonly used T2 weighted (T2w) fat suppressed images [3] do not allow for confident distinction between axons and accompanying vessels. However, 3D reversed fast imaging with steady state free procession sequence (PSIF) suppresses the signal form surrounding vessels [4], generating nerve-selective images. Regarding diffusion-weighted imaging of the mandibular nerves, echo planar imaging (EPI) is limited due to susceptibility artifacts in the oral cavity. Readout-segmented EPI (rs-EPI) is less prone to artifacts [5] but requires longer scan durations. However, new simultaneous multi-slice (SMS) acquisition technologies with blipped Controlled Aliasing In Parallel Imaging (CAIPI) may allow scan durations adequate for clinical routine [6]. The aim of this study was to propose a multiparametric approach for “MR neurographic orthopantomograms” for gathering comprehensive information about the mandible, teeth and nerves in a single examination.

IRB approved study. Ten asymptomatic volunteers (7 women, age (mean±SD) 28.5±9.1 years) and 3 men (age 26.6±0.6 years) were imaged at 3.0 Tesla (Skyra, Siemens Healthcare, Erlangen, Germany) using a 64-channel head coil.

Imaging protocol. For ultra-short TE bone imaging, a 3D-PETRA (Pointwise Encoding Time reduction with Radial Acquisition) single-echo sequence (minimal TE, 0.07ms, see Fig.1) was applied [7]. For morphological nerve imaging, a 3D-PSIF sequence with vascular signal suppression was used. Additionally, a T2w SPACE (Sampling Perfection with Application optimized Contrasts using different flip angle Evolution) sequence with STIR (short tau inversion recovery) fat suppression was acquired. For diffusion imaging of the mandibular nerve, a sequence based on rs-EPI with simultaneous multi-slice excitation (SMS) was applied. Additional to the conventional rs-EPI, sequences were applied with two- and three-fold (2/3xSMS) slice acceleration for 8 volunteers. All diffusion sequences were performed twice for SNR calculation.

Image analysis. Image analysis was performed on a syngo-via platform (Version VB10A, Neuro3D and Frontier cinematic rendering, Siemens Healthcare) by two independent readers. For qualitative assessment, overall image quality, delineation of the mandibular canal/nerve and artifact scores were rated on a 4-point Likert-scale ranging from 1 (excellent) to 4 (non-diagnostic). For quantitative assessment of the diffusion data, tractography of both mandibular nerves was performed. The following values were extracted: Number of generated tracts, fractional anisotropy (FA), mean diffusivity (MD). SNR was calculated according to [6]: $$\frac{SI\times\sqrt{2}}{\sigma}$$ (SI, signal intensity on b=1000; σ, standard deviation of the signal intensity on the subtraction image). For inter-reader agreement, Kappa statistics were calculated. Results were compared using ANOVA, paired-sample t-tests and Wilcoxon signed-rank tests.

Qualitative analysis. Inter-rater agreement ranged from “substantial” to “almost perfect” (kappa, 0.602–0.815). All sequences yielded excellent image quality (mean±SD; PETRA, 1.20±0.42; T2w SPACE STIR, 1.30±0.48; PSIF, 1.20±0.42; rs-EPI, 1.40±0.70), excellent delineation of the mandibular canal/nerve (mean±SD, PETRA, 1.10±0.32; T2w SPACE STIR, 1.30±0.48; PSIF, 1.20±0.42, rs-EPI, 1.40±0.70) and no or low distorsion artifacts (mean±SD, PETRA, 1.40±0.52; T2w SPACE STIR, 1.20±0.42; PSIF, 1.30±0.48; rs-EPI, 1.30±0.48, see Fig. 2 and 3). However, PETRA and PSIF yielded moderate ghosting/motion artifacts (mean±SD; PETRA, 2.50±0.53; PSIF, 2.30±0.48).

Quantitative diffusion analysis. Tractography was feasible in all volunteers for all sequences except for one volunteer using 3xSMS rs-EPI (Fig. 4). Quantitative analysis yielded following values for rs-EPI (mean±SD): number of generated tracts, 52.73±43.60; FA, 0.43±0.07; MD (mm2/s), 0.0014 ±0.000. Analysis between different accelerations revealed no significant difference regarding number of tracts, FA, MD or SNR between rs-EPI and 2xSMS rs-EPI. However, 3xSMS rs-EPI yielded significantly lower SNR (p=0.014, Fig. 5), number of tracts (p=0.030) and MD (p=0.002) compared to rs-EPI/2xSMS rs-EPI.3D-PETRA generated robust images of the mandibular canal. 3D-PSIF with vascular signal suppression enabled selective visualization of the mandibular nerve. The rs-EPI sequence elicited no major distortion artifacts due to field inhomogeneities in the oral cavity. Results suggest that a two-fold acceleration provides robust results, while three-fold acceleration yielded lower SNR, number of tracts and MD compared to standard and two-fold accelerated rs-EPI.The proposed technique of ‘MR neurographic orthopantomogram’ exploiting ultra-short TE imaging complemented with selective morphological and accelerated diffusion weighted neurography was feasible and allowed comprehensive assessment of osseous texture and neural micro-architecture in a single examination.