Diffusion-Weighted MR Neurography Of Extremity Nerves At3.0T

Part I Diffusion-weighted MR Neurography of the Tibial and Common Peroneal Nerves with Different Motion Probing Gradients at3.0TPurpose:To prospectively and intraindividually compare the image quality of diffusion-weighted MR neurography (DW-MRN) of the tibial and common peroneal nerves obtained using different motion probing gradients (MPGs) at3.0T.Materials and Methods:Total21healthy volunteers underwent DW-MRN at the level of knee (unilateral imaging) at a3.0T magnetic resonance system with anterior-posterior directional MPGs, right-left directional MPGs, three directional MPGs and six directional MPGs. Post-processing included reformatting of3D volume images from the entire axial source DW-MRN images on Philips Extended MR Workspace. This was performed using the full volume MIP reconstruction and volume editing which allowed cutting out of the superimposed structures with similar hyperintensity (such as veins and articular fluids) in order to optimize visualization of the nerves. This method allowed3D rotations which generally consisted of15volume images processed around the longitudinal axis. Additional oblique planes were reformatted and gray-scale inversion was applied when necessary. These post-processed3D volume MIP images were referred to as DW-MRN images. The apparent SNR and CNR of the tibial and common peroneal nerves were calculated on axial source DW-MRN images. DW-MRN images were blindly evaluated by two radiologists using a four-point grading scale:1=poor (the nerve was partially visible); 2=moderate (the entire nerve was visible and of moderate signal intensity);3=good (the entire nerve was visible and of good signal intensity);4=excellent (the entire nerve was visible and of excellent signal intensity). Significance was determined by using Friedman and paired Wilcoxon tests. Interobserver agreement for visualization scores was calculated by using the Kappa statistic. The level of agreement was defined as follows:values of0-0.20were considered to indicate poor agreement;0.21-0.40, fair agreement;0.41-0.60, moderate agreement;0.61-0.80, good agreement; and0.81-1.00, excellent agreement.Results:The tibial nerve image qualityThe SNR of tibial nerves on DW-MRN with anterior-posterior directional MPGs, right-left directional MPGs, three directional MPGs and six directional MPGs were4.109±0.527,4.31±0.822,3.429±0.495and3.300±0.469respectively; CNR were0.604±0.042,0.606±0.133.0.543±0.052and0.529±0.054. The mean scores for tibial nerve image quality were3.714±0.561,3.762±0.436,2.619±0.590and2.048±0.590;3.667±0.577,3.714±0.463,2.571±0.507and1.714±0.463for observers A and B, respectively. Interobserver agreement was good and the κ values were0.691. DW-MRN of tibial nerves with unidirectional MPGs was better than with three directional MPGs and six directional MPGs (P<0.05). DW-MRN of tibial nerves with three directional MPGs was better than with six directional MPGs (P<0.05). The SNR and CNR of tibial nerves on DW-MRN were better with right-left unidirectional MPGs than with anterior-posterior unidirectional MPGs (P<0.05), but the differences of SNR and CNR between them were0.202and0.002only and there were no significant differences between them on the grading scales.The common peroneal nerve image qualityThe SNR of common peroneal nerves on DW-MRN with anterior-posterior directional MPGs, right-left directional MPGs, three directional MPGs and six directional MPGs were3.060±0.395,3.085±0.4444,2.888±1.474and2.628±0.371respectively; CNR were0.503±0.048,0.505±0.053,0.459±0.086and0.443±0.058. The mean scores for common peroneal nerve image quality were3.476±0.680.3.57±0.676.2.000±0.548and1.857±0.478;3.524±0.602.3.571±0.598.2.048±0.384and1.905±0.301for observers A and B. respectively. Interobserver agreement was good and the κ values were0.699. DW-MRN of common peroneal nerves with unidirectional MPGs was better than with three directional MPGs and six directional MPGs (P<0.05). There were no significant differences between DW-MRN with right-left and anterior-posterior unidirectional MPGs (P>0.05), and there were no significant differences between DW-MRN with three directional MPGs and six directional MPGs (P>0.05).Conclusion:Our preliminary results suggest that DW-MRN is feasible for providing three-dimensional visualization of the tibial and common peroneal nerves, and it should be perfomied with anterior-posterior or right-left unidirectional MPGs at3.0T.Part II Diffusion-Weighted MR Neurography of Extremity Nerves With Unidirectional Motion-Probing Gradients at3.0T: Feasibility StudyPurpose:To demonstrate the feasibility and performance of diffusion-weighted MR neurography (DW-MRN) of extremity nerves with unidirectional motion-probing gradients (MPGs) at3.0T.Materials and Methods:Institutional review board approval and informed consent were obtained.39healthy volunteers(mean age.31.1±9.5years [standard deviation]: age range,24-61years;20males:19females; upper arm.7; elbow.7; forearm.7; thigh.9:calf,9) and14patients (32.2±15.9years:age range,6-61years;10males;4females) underwent DW-MRN of extremity nerves with unidirectional motion-probing gradients (anterior-posterior direction only) on a3.0T MR unit. The volunteers had no prior history or findings related to nerve pathology at the time of the study. Follow-up examinations were performed in3patients examined at2months, 5months and1year after surgery respectively. All patients also underwent conventional MR imaging including axial or coronal T1-weighted imaging, coronal T2-weighted imaging, and axial or coronal T2-weighted spectrally adiabatic inversion recovery imaging (T2WI-SPAIR).The MPGs were implemented in anterior-posterior direction. Other imaging parameters for DW-MRN with the2-channel large-flex coil and8-channel knee coil were similar as in the part one. The imaging parameters for DW-MRN with the16-channel torso coil and NV coil were similar, except for field of view of30×30cm, acquisition matrix of112×109,60-80slices acquired, echo-planar imaging factor of47, and sensitivity-encoding factor of2.5. For bilateral imaging of lower extremities in cases with neurofibromatosis, the16-channel torso coil was used at the knee and calf levels.Post-processing of DW-MRN images was similar as in the part one. Two radiologists blindly and independently evaluated the imaging quality of extremity nerves of volunteers on DW-MRN images using a four-point grading scale as in the part one. For patients, the same two radiologists blindly and independently evaluated the DW-MRN and T2WI-SPAIR images. The DW-MRN and T2WI-SPAIR images were separately assessed. The extent of the lesions and the relationship with adjacent nerves was assed using a three-point grading scale:1=The extent of the lesions and the relationship with adjacent nerves identified with certainty;2=Probably identified or not all lesions were identified;3=Not identified.The statistical analysis was performed with SPSS software. Interobserver agreement for visualization scores was calculated by using the kappa statistic as in the part one. Paired Mann-whitney U-test was used to assess differences in conspicuity scores between DW-MRN and T2WI-SPAIR images of the same patient. The statistically significant difference level was set at P<0.05.Results:Volunteer StudyThe long trajectories of major extremity nerves, including radial, median, ulnar, sciatic and tibial nerves, could be easily visualized on DW-MRN images, and the mean scores for nerve image quality were3.79＋0.43、3.81±0.51、3.43±0.81、4±0、2.56=1.13and3.86±0.36、3.86±0.48、3.38±0.74、4±0、2.67±1.12for observers A and B, respectively. The mean scores of all nerves were3.61±0.75(standard deviation), and3.64±0.71for observers1and2, respectively. The interobserver agreement was good (κ=0.766).In2(22%) of9volunteers examined at the level of calf, the tibial nerves were poorly visualized due to the overlap of adjacent vascular hyperintensity, especially at the distal calf. The veins traveled alongside the nerves were well suppressed on DW-MRN images in this study, except at the level of calf. The vessels displayed a fluctuating low to high signal intensity along their length and the tibial nerves displayed almost uniformly moderate or high signal intensity. However, differentiation between them was not always successful when the vessels traveled alongside the tibial nerve. These vessels rendered the discrimination difficult and could not be removed in post-processing.Patient studyIn patients, there was no discrepancy between observers (κ=1). The extent of lesions and the relationship with adjacent nerves were clearly and easily depicted on DW-MRN images in all patients with a comprehensive overview. Except in2patients after surgery (score=1), the scores of all patients were3on DW-MRN images. On T2WI-SPAIR images, the scores were3in only7patients. The DW-MRN images provided improved identification of the extent of the lesions and the relationship with adjacent nerves compared with the T2WI-SPAIR images (P<0.05).In the5patients with schwannoma and the patient with neurofibroma. ovoid tumors and their original nerves were clearly displayed on DW-MRN images, but differentiation between schwannoma and neurofibroma was still difficult. In follow-up examination of one patient with tibial nerve schwannoma at2months after surgery, no recurrence was detected and the tibial nerve was poorly distinguishable from suiTounding tissues with hyperintensity in the operation area. In the patients with neurofibromatosis, the type (discrete or plexiform), number, size and distribution of tumors in extremities were clearly displayed on DW-MRN straightforwardly. In the patient with aggressive fibromatosis, the irregular tumor with hyperintensity was seen immediately adjacent to the left sciatic nerve, but a distinct boundary between them was clearly displayed with3D rotations on DW-MRN images. So displacement but not interruption of the nerve could be demonstrated. In the follow-up examination at5months after surgery, no recurrence was detected and the left sciatic nerve was visualized in a relatively normal course.In patients with torsional nerve injury, DW-MRN images of the arms showed diffuse enlargement and multiple "hourglass-like" constrictions of the radial or median nerves, with a "sausage-like" appearance. In the patient with partially sciatic nerve injury, the extent of asymmetric hyperintensity and enlargement of the left sciatic nerve without discontinuity as compared to normal-appearing right sciatic nerve could be clearly detected.ConclusionThis preliminary study demonstrates that3.0T DW-MRN with unidirectional motion-probing gradients is feasible for providing three-dimensional visualization of major extremity nerves, including radial, median, ulnar, sciatic and tibial nerves. Based on the functional diffusion characteristics, DW-MRN can depict the location and extent of lesions, and the anatomic relationship to the adjacent nerves with a comprehensive overview.