- Email: [email protected]
Greater occipital nerve MR tractography: Feasibility and anatomical considerations
G Model NEURAD-668; No. of Pages 5
ARTICLE IN PRESS Journal of Neuroradiology xxx (2017) xxx–xxx
Available online at
ScienceDirect www.sciencedirect.com
Original Article
Greater occipital nerve MR tractography: Feasibility and anatomical considerations Adrian Kastler a,b,∗ , Arnaud Attye a,b , Olivier Heck a,b , Florence Tahon a , Kamel Boubagra a , Irène Tropes b , Sylvie Grand a,b , Alexandre Krainik a,b a b
Neuroradiology and MRI Unit, CS 10217, Grenoble Alpes University, 38043 Grenoble cedex 9, France Grenoble Alpes University, IRMaGe, 38000 Grenoble, France
a r t i c l e
i n f o
Article history: Available online xxx Keywords: Greater occipital nerve Neuralgia Tractography MR neurography MRI
a b s t r a c t Background and purpose. – To assess the feasibility of greater occipital nerve (GON) tractography using a fully automated tractography technique on the whole-neck volume, in comparison with anatomical knowledge. Methods. – Healthy subjects were consecutively included in this study if they had no history or symptoms of headache or brain disorder. A 3T MRI scanner with a 32 channel head coil was used. The following parameters for Diffusion Weighed (DWI) were used: b value of 1000 s/mm2 , 32 directions, acquired voxel size: 2 mm isotropic. High-Order tractography with the Constrained Spherical Deconvolution (CSD) model was generated. Track-Weighted Imaging (TWI) maps were generated with MRTrix. Two radiologists performed blind evaluations of the GON pathways on TWI maps. Results. – A total of 20 healthy subjects were included (12 males and eight females, mean age 53.8 years old). In comparison with anatomical atlas, GON complete visualization (from C1–C2 origin to muscular emergence) was possible in 18 out of 20 healthy subjects. In two cases, GON was not visible in the cervical spine foramen. Conclusion. – Tractography through TWI is a feasible technique to accurately depict GON. This technique may appear as a promising technique for therapeutic management of patients with occipital neuralgia. © 2017 Elsevier Masson SAS. All rights reserved.
Introduction Diffusion tensor imaging (DTI), based on tractography, has been shown to be useful in animal models to monitor peripheral nerve degeneration [1] or for the assessment of peripheral nerve infiltration by malignant tumors in human beings [2] [3]. Yet, DTI is known to potentially yield misleading information regarding the actual pathways of white matter in the brain [4]. The constrained spherical ceconvolution (CSD) method gives an estimation of fiber orientation distribution, directly from diffusionweighted MRI (DWI) data, without the need for prior assumptions
Abbreviations: GON, greater occipital nerve; MRI, magnetic resonance imaging; CT, computed tomography; DTI, diffusion tensor imaging; DWI, diffusion weighted imaging; TWI, track weighted imaging; CSD, constrained spherical deconvolution; FA, fractional anisotropy; ODF, orientation density fiber. ∗ Corresponding author. Neuroradiology and MRI Unit, CS 10217, Grenoble Alpes University, 38043, Grenoble cedex 9, France. E-mail address: [email protected] (A. Kastler).
regarding the number of fiber populations present [5]. This model has produced super-resolution tractography with the so-called track weighted imaging (TWI) [6] providing high anatomical contrast, by incorporating additional information from diffusion tractography modeling. This method was recently used to show the feasibility of intraparotid facial nerve [7] tractography, to identify nerve contact with parotid tumors. The greater occipital nerve (GON) has been shown to be implicated in the management of various types of cranio-facial pain syndromes [8,9]. The anatomical course of the GON in the deep cervical spaces has been described with cadaver studies [10] (Fig. 1): its origin arises from the C2 dorsal ramus between the atlas and the axis (first segment, S1), it then curves around the inferior border of the oblique inferior muscle (first bend, B1), then courses between the inferior oblique and semi-spinalis muscles (second segment, S2) before perforating the semi-spinalis and trapezius muscles (second bend, B2) and emerges subcutaneously in its third segment (S3) in the sub-occipital region. The deep cervical space course of the GON has never been assessed with MRI tractography. Therefore, the objective of this study was to assess the feasibility of GON MRI diffusion tractography at 3 T, in comparison with the anatomical atlas.
http://dx.doi.org/10.1016/j.neurad.2017.09.001 0150-9861/© 2017 Elsevier Masson SAS. All rights reserved.
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
2
Fig. 1. Greater occipital nerve anatomical description. The image on the left is an anatomical drawing representing the course of the GON. It arises from the posterior ramus of the C2 root through the C1–C2 foramen (origin), and then presents three segments (S1, S2 and S3) and two bends (B1 an B2). Three areas of vulnerability have been described: at it’s origin, first bend and emergence.The image on the left shows a ROI based tractogram of the GON, accurately depicting the whole course of the GON in the deep cervical spaces. Note that the acquisition is limited from the origin to the emergence, S3 is not seen.
Material and methods Subjects and data acquisition The study protocol was approved by our institutional review board for a prospective study in healthy subjects (IRB 5891). Healthy subjects aged 18 years or over were consecutively included in this study if they had: • no history or symptoms of brain or psychiatric disorder and no medical treatment; • undergone DWI and post-processing TWI. Two radiologists managed the subjects who required MR scans. ® We performed the scan on a 3 T MR imaging Philips ACHIEVA 3.0 T TX with a 32 channel head coil. The MR scan included transverse, sagittal and coronal T2weighted spin-echo sequences. The parameters of the diffusion sequence were: b value 1000 s/mm2 ; 32 directions; acquired voxel size: 2 mm isotropic; field of view 220 mm; single-shot spin-echo sequence; scan duration: 15 . An additional pair of b = 0 images (one with reversed phase encoding) was acquired for estimation of the inhomogeneity field [11]. Pre-processing and tractography reconstructions Diffusion image data (Fig. 2A) were corrected for B0 field in homogeneities and eddy current using top-up and eddy ® tools in FSL5 . The pre-processing steps were performed using MRtrix package software (JD Tournier, Brain Research Institute, Melbourne, Australia, http://www.brain.org.au/software/). Furthermore, for fiber orientation distribution estimation using the CSD model, the number of spherical harmonic terms was set to six. Whole-neck streamline tractography with 10 million streamlines
was then generated. Initial fractional anisotropy (FA) cut-off for terminating tracks was set to 0.3, an isotropic voxel size of 300 m and the minimum length of fibers to 20 mm as previously described in another cranial nerve tractography study [7]. To ensure that quantitative values yielded from DWI data were estimated in accordance with the SNR of the neck region, we excluded voxels with negative eigen-values before subsequent post-processing. TWI maps were performed (Fig. 2B) to produce an image of the mean length of fibers through each voxel rather than the count [12]. The post-processing lasted around 15 minutes using a com® puter with a multi-core processor (Intel Core i7, 3 GHz, Intel ® Corporation , USA). TWI maps analyses and GON study Two radiologists evaluated MRI tractography data including TWI and tractography reconstructions. The whole TWI tractography volume was used to assess GON visualization (Fig. 2B). For qualitative analyses, we visually evaluated the intensity of the GON with a three-grade ranking system. Each GON was assigned the following scores: • • • •
Score 0: no GON visualization; Score 1: Partial GON visualization; Score 2: Complete GON visualization; Score 2 was considered to represent physiological features. Scores 0 and 1 (Fig. 2) were considered as TWI failure.
The results of MR imaging studies of each patient were considered positive or negative for GON visualization by each observer. For an MR imaging–positive result, the GON had to be assigned a score of 2. Moreover, the following criteria were also assessed: position of B1 compared to the C2 pedicle (below or above lower aspect of
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
3
Fig. 2. Steps of greater occipital nerve tractography process with MRtrix. A. Raw DWI image at the C1–C2 foramen level after susceptibility and movement artifacts correction. B. Example of TWI mapping at the same C1–C2 level. This map was obtained after a first tractography process with generation of 10 millions fibers on the whole neck volume, then automatically converted into this picture file. High signal intensity (red) is correlated to the mean length of the tracted fiber. The direct visualization of GONs (white arrow heads) allow to ease extraction of coordinates before placing ROI. C. Bilateral GON (white arrow heads) tractogram after ROI extraction superimposed on morphological axial T2 weighted Imaging. D. Selective TWI GON representation (white arrow heads), with axial T2 superimposition. Finally, the track files were converted into GON picture files, with super-resolution technique, which allows GON representation with an isotropic resolution of 0.3 mm. This technique is useful for exploring the GON course in 3D slice by slice.
pedicle); intra- or extra-muscular course of B1 (through the inferior oblique muscle); visualization of motor branches; anatomical variations of the S2 segment of the GON. Inter-rater agreement on detecting GON was estimated using Cohen’s kappa coefficient. In addition, FA of the GON on both sides was assessed, using FA maps as a template as well as two regions of interest to extract the GON from TWI maps (Fig. 2C). A Student’s t-test was performed to assess the differences in FA between the two GON among subjects. Continuous statistical data were analyzed using SPSS software v22.0 (IBM, Inc., Armonk, New York, USA).
whether the course of B1 was situated intra- or extra-muscularly to the inferior oblique muscle in all cases. Motor branches were identified in only four cases. GON S2 segment assessment showed no variations in all 18 cases. Quantitative method by FA measurement GON fractional anisotropy was estimated to 0.47 ± 0.08 in subjects. There was no significant difference in FA measurements between the two GON in each diffusion acquisition. Discussion
Results Anatomical and clinical considerations GON visualization A total of 20 subjects were included (12 males and 8 females, mean age 53.8 yr). In comparison with the anatomical atlas, GON complete visualization (score 2) was possible in 18 out of 20 subjects. In two cases, GON was not visible in the cervical spine foramen (score 1). No subjects referred with score 0. Inter-rater agreement on detecting GON was estimated as being 1.0 on 2D reconstructions, implying that both radiologists were able to systematically detect GONs among the 18 subjects (kappa test), without discordance for the two remaining cases. In all but four cases, the first bend (B1) was depicted above the inferior aspect of the ipsilateral pedicle. It was not possible to assess
The anatomical course of the GON has been widely studied due to its implication in various types of headache syndromes [8,9]. The GON has been described to have three segments and two bends [13] (Fig. 1) with several possible areas of GON vulnerability [14]: first, at its origin from the C2 dorsal ramus between the atlas and the axis; second, at the first bend where the GON curves around the inferior oblique capitis muscle; third, at its superficial emergence when perforating the aponeurosis of the trapezius muscle. Anatomical studies of the GON have mainly focused on assessing the superficial emergence of the GON through the trapezius muscle, because this vulnerability site is the most used target site for GON infiltration in clinical practice [15–17] based on anatomical
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5 4
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
landmarks. The anatomical variability of the GON seems to be greater after the second bend and throughout the third segment of the GON [13,14,18]. However, only a few anatomical studies have assessed the initial segments of the GON and its variability [19]. Our results show that the GON is correctly and accurately depicted with TWI tractography. Moreover, our study showed the existence of subtle anatomical variations of the GON course in its initial bend in comparison with the C2 pedicle, however, no variations were found in the second segment of the GON (S2), during its course between the inferior oblique and capitis muscles. Previous studies have targeted the initial course of the GON (S1, B2 and S2) with satisfactory results [20,21]. Therefore, the presence of anatomical variations may be of interest to the interventional physician targeting the GON in some procedures, such as GON neurolysis, which has also been described [14,22]. In these specific situations, TWI tractography may appear as a useful pre-planning tool allowing detection of both possible anatomical variations, and accurate nerve localization as neurolysis procedures require immediate proximity between the needle and the nerve [23] for a successful outcome. In two cases, the GON was not seen in the C1–C2 foramina. DWI is prone to distortion and motion artifacts, which may be problematic for the study of GON in the foramina. However, to limit this problem we associated a method of diffusion optimization specifically adapted to correct for patient movements with global track-weighted imaging. GON motor branches were only detected in four of 18 cases. This might be explained by anatomical variability [14] of the motor branches. There are also technical limitations for depicting small nerves because the CSD model relies on probabilistic algorithms thatgive priority to large nerve trunk detection over small branches. We can also mention that CSD has proved to be valuable in overcoming issues of intraparotid VIIn variability and complex branching patterns. Moreover, CSD is based on orientation density fiber (ODF) reconstructions, known to be superior to diffusion ODF methods [19], in terms of tractography accuracy. Track-weighted imaging, a new qualitative tool to assess nervous pathways The TWI technique was recently introduced as a qualitative tractography method due to its high anatomical contrast [6]. In this acquisition, the nervous pathways can be combined with an anatomical reference map to facilitate nerve identification [7]. Whilst seed-based tractography is generally user-dependent, requiring in-depth structural neuroanatomical knowledge, a whole cervical fiber tracking approach offers a user-independent method for extracting visual information from the tractogram itself. The interest of FA measures in comparison with direct visual assessment of GON may be discussed. Quantitative FA analysis is currently the gold standard in tractography studies; TWI map does not yet appear as an ideal tool to provide quantitative values [24]. In addition, this study demonstrated no significant difference between GON FA values on both sides. It may be interesting to extract FA in patients with neuralgia, looking for FA decrease on the affected nerve in comparison with the contralateral side. Yet, the use of regions of interest (ROI) required by the FA method may be difficult in clinical practice and user-dependent, as ROI placement requires precise anatomical knowledge of GON. With this regard, it seems interesting to highlight the absence of GON discontinuity in 18 out of 20 subjects. Calamante et al. have previously validated TWI maps using direct comparison with mouse brain histological patterns [12]. In this work, the bright areas on the TWI maps significantly matched the dark areas in the myelin stained images. In another TWI study in patients with glioblastoma, higher relative TWI values of the tumor location yielded a higher likelihood of increased tumor proliferation, greater architectural
disruption, and microvascular hyperplasia [25]. Previous studies have shown the feasibility of such a technique on retinal tractography in healthy subjects [26]. In this GON study, we were unable to depict differences in the signal intensity of the nerve in the different subjects. However, future studies may depict modification of the GON in pathological conditions such as greater occipital neuralgia or other cranio-facial pain syndromes. Finally, and concerning the feasibility of TWI mapping in clinical practice, the acquisition time does not represent an issue, as the sequence used in this study lasts less than 10 minutes. Therefore, the main challenge is the possible movement artifact that may occur in some patients, although most artifacts can be dealt with in the post-processing steps. Post-processing is also a challenge in the everyday work flow, as 20 minutes are needed to edit the TWI maps. Limitations of this study Limitations of the study include the relatively small number of subjects included. Moreover, previous high-order tractography studies recommended the acquisition of more than 45 DW directions to avoid uniformity problems of diffusion weighted gradient directions, and to meet SNR requirements. However, this study showed successful identification of the GON in 18 out of 20 subjects. The method used is also user-dependent for FA extraction since it required anatomical knowledge concerning the GON course. However, the “Whole-Neck” streamlines approach with generation of tractograms in the neck area allowed an efficient identification of the GON course. In addition, the effect of patient age on FA measurement could be further assessed. Conclusion MRI diffusion tractography of the GON is feasible, allowing accurate depiction of the GON and may be applied in clinical practice. The usefulness of GON tractography in clinical routine is yet to be demonstrated. However, it may be very interesting in patients suffering from GON-mediated cranio-facial pain, in order to detect possible GON impairment. It may also appear as a potential preplanning tool in specific procedures targeting the GON. GON MRI diffusion tractography will need further investigations. Disclosure of interest The authors declare that they have no competing interest. References [1] Takagi T, Nakamura M, Yamada M, Hikishima K, Momoshima S, Fujiyoshi K, et al. Visualization of peripheral nerve degeneration and regeneration: monitoring with diffusion tensor tractography. Neuroimage 2009;44:884–92. [2] Kasprian G, Amann G, Panotopoulos J, Schmidt M, Dominkus M, Trattnig S, et al. Peripheral nerve tractography in soft tissue tumors: a preliminary 3-tesla diffusion tensor magnetic resonance imaging study. Muscle Nerve 2015;51:338–45. [3] Rouchy RC, Attyé A, Troprès I, Medici M, Kastler A, Righini C, Krainik A. Facial nerve tractography: a new tool to detect perineural invasion in parotid cancers. 10.1016/j.neurad.2017.01.005. [4] Farquharson S, Tournier JD, Calamante F, Fabinyi G, Schneider-Kolsky M, Jackson GD, et al. White matter fiber tractography: why we need to move beyond DTI. J Neurosurg 2013;118:1367–77. [5] Tournier JD, Calamante F, Gadian DG, Connelly A. Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution. Neuroimage 2004;23:1176–85. [6] Calamante F, Tournier JD, Kurniawan ND, Yang Z, Gyengesi E, Galloway GJ, et al. Super-resolution track-density imaging studies of mouse brain: comparison to histology. Neuroimage 2012;59:286–96. [7] Attye A, Karkas A, Tropres I, Roustit M, Kastler A, Bettega G, et al. Parotid gland tumours: MR tractography to assess contact with the facial nerve. European radiology 2015, http://dx.doi.org/10.1007/s00330-015-4049-9. [8] Anthony M. Headache and the greater occipital nerve. Clin Neurol Neurosurg 1992;94:297–301.
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
[9] Blumenfeld A, Ashkenazi A, Napchan U, Bender SD, Klein BC, Berliner R, et al. Expert consensus recommendations for the performance of peripheral nerve blocks for headaches-a narrative review. Headache 2013;53:437–46. [10] Loukas M, El-Sedfy A, Tubbs RS, Louis Jr RG, Wartmann CH, Curry B, et al. Identification of greater occipital nerve landmarks for the treatment of occipital neuralgia. Folia Morphol (Warsz) 2006;65:337–42. [11] Andersson JL, Skare S, Ashburner J. How to correct susceptibility distortions in spin-echo echo-planar images: application to diffusion tensor imaging. Neuroimage 2003;20:870–88. [12] Pannek K, Mathias JL, Bigler ED, Brown G, Taylor JD, Rose SE. The average pathlength map: a diffusion MRI tractography-derived index for studying brain pathology. Neuroimage 2011;55:133–41. [13] Vital JM, Grenier F, Dautheribes M, Baspeyre H, Lavignolle B, Senegas J. An anatomic and dynamic study of the greater occipital nerve (n. of Arnold). Applications to the treatment of Arnold’s neuralgia. Surg Radiol Anat 1989;11:205–10. [14] Bogduk N. The anatomy of occipital neuralgia. Clin Exp Neurol 1981;17:167–84. [15] Dach F, Eckeli AL, Ferreira Kdos S, Speciali JG. Nerve block for the treatment of headaches and cranial neuralgias - a practical approach. Headache 2015;55(1):59–71. [16] Ambrosini A, Vandenheede M, Rossi P, Aloj F, Sauli E, Pierelli F, et al. Suboccipital injection with a mixture of rapid- and long-acting steroids in cluster headache: a double-blind placebo-controlled study. Pain 2005;118:92–6. [17] Weibelt S, Andress-Rothrock D, King W, Rothrock J. Suboccipital nerve blocks for suppression of chronic migraine: safety, efficacy, and predictors of outcome. Headache 2010;50:1041–4.
5
[18] Bovim G, Bonamico L, Fredriksen TA, Lindboe CF, Stolt-Nielsen A, Sjaastad O. Topographic variations in the peripheral course of the greater occipital nerve. Autopsy study with clinical correlations. Spine 1991;16:475–8. [19] Tubbs RS, Watanabe K, Loukas M, Cohen-Gadol AA. The intramuscular course of the greater occipital nerve: novel findings with potential implications for operative interventions and occipital neuralgia. Surg Neurol Int 2014;5:155. [20] Pougnard-Bellec F, Rolland Y, Morel D, Meadeb J, Marin F, Duvauferrier R. [Efficacy of C1-C2 block with posterior parasagittal approach in the treatment of Arnold neuralgia in 24 patients]. J Radiol 2002;83:133–9. [21] Aubry S, Kastler B, Bier V, Hadjidekov V, Hussein HH, Fergane B. Evaluation of the effectiveness of CT-guided infiltration in the treatment of Arnold’s neuralgia. Neuroradiology 2009;51:163–8. [22] Cohen SP, Peterlin BL, Fulton L, Neely ET, Kurihara C, Gupta A, et al. Randomized, double-blind, comparative-effectiveness study comparing pulsed radiofrequency to steroid injections for occipital neuralgia or migraine with occipital nerve tenderness. Pain 2015;156:2585–94. [23] Bogduk N, Dreyfuss P, Govind J. A narrative review of lumbar medial branch neurotomy for the treatment of back pain. Pain Med 2009;10:1035–45. [24] Jones DK, Knosche TR, Turner R. White matter integrity, fiber count, and other fallacies: the do’s and don’ts of diffusion MRI. Neuroimage 2013;73:239–54. [25] Barajas Jr RF, Hess CP, Phillips JJ, Von Morze CJ, Yu JP, Chang SM, et al. Superresolution track density imaging of glioblastoma: histopathologic correlation. AJNR Am J Neuroradiol 2013;34:1319–25. [26] Jean C, Attyé A, Grand S, Lecler A, Chiquet C, Aptel F, Lamalle L, Krainik A. Retinal tractography: influence of diffusion acquisition, ocular laterality and ocular dominance in healthy subjects. 10.1016/j.neurad.2017.01.035.
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
ARTICLE IN PRESS Journal of Neuroradiology xxx (2017) xxx–xxx
Available online at
ScienceDirect www.sciencedirect.com
Original Article
Greater occipital nerve MR tractography: Feasibility and anatomical considerations Adrian Kastler a,b,∗ , Arnaud Attye a,b , Olivier Heck a,b , Florence Tahon a , Kamel Boubagra a , Irène Tropes b , Sylvie Grand a,b , Alexandre Krainik a,b a b
Neuroradiology and MRI Unit, CS 10217, Grenoble Alpes University, 38043 Grenoble cedex 9, France Grenoble Alpes University, IRMaGe, 38000 Grenoble, France
a r t i c l e
i n f o
Article history: Available online xxx Keywords: Greater occipital nerve Neuralgia Tractography MR neurography MRI
a b s t r a c t Background and purpose. – To assess the feasibility of greater occipital nerve (GON) tractography using a fully automated tractography technique on the whole-neck volume, in comparison with anatomical knowledge. Methods. – Healthy subjects were consecutively included in this study if they had no history or symptoms of headache or brain disorder. A 3T MRI scanner with a 32 channel head coil was used. The following parameters for Diffusion Weighed (DWI) were used: b value of 1000 s/mm2 , 32 directions, acquired voxel size: 2 mm isotropic. High-Order tractography with the Constrained Spherical Deconvolution (CSD) model was generated. Track-Weighted Imaging (TWI) maps were generated with MRTrix. Two radiologists performed blind evaluations of the GON pathways on TWI maps. Results. – A total of 20 healthy subjects were included (12 males and eight females, mean age 53.8 years old). In comparison with anatomical atlas, GON complete visualization (from C1–C2 origin to muscular emergence) was possible in 18 out of 20 healthy subjects. In two cases, GON was not visible in the cervical spine foramen. Conclusion. – Tractography through TWI is a feasible technique to accurately depict GON. This technique may appear as a promising technique for therapeutic management of patients with occipital neuralgia. © 2017 Elsevier Masson SAS. All rights reserved.
Introduction Diffusion tensor imaging (DTI), based on tractography, has been shown to be useful in animal models to monitor peripheral nerve degeneration [1] or for the assessment of peripheral nerve infiltration by malignant tumors in human beings [2] [3]. Yet, DTI is known to potentially yield misleading information regarding the actual pathways of white matter in the brain [4]. The constrained spherical ceconvolution (CSD) method gives an estimation of fiber orientation distribution, directly from diffusionweighted MRI (DWI) data, without the need for prior assumptions
Abbreviations: GON, greater occipital nerve; MRI, magnetic resonance imaging; CT, computed tomography; DTI, diffusion tensor imaging; DWI, diffusion weighted imaging; TWI, track weighted imaging; CSD, constrained spherical deconvolution; FA, fractional anisotropy; ODF, orientation density fiber. ∗ Corresponding author. Neuroradiology and MRI Unit, CS 10217, Grenoble Alpes University, 38043, Grenoble cedex 9, France. E-mail address: [email protected] (A. Kastler).
regarding the number of fiber populations present [5]. This model has produced super-resolution tractography with the so-called track weighted imaging (TWI) [6] providing high anatomical contrast, by incorporating additional information from diffusion tractography modeling. This method was recently used to show the feasibility of intraparotid facial nerve [7] tractography, to identify nerve contact with parotid tumors. The greater occipital nerve (GON) has been shown to be implicated in the management of various types of cranio-facial pain syndromes [8,9]. The anatomical course of the GON in the deep cervical spaces has been described with cadaver studies [10] (Fig. 1): its origin arises from the C2 dorsal ramus between the atlas and the axis (first segment, S1), it then curves around the inferior border of the oblique inferior muscle (first bend, B1), then courses between the inferior oblique and semi-spinalis muscles (second segment, S2) before perforating the semi-spinalis and trapezius muscles (second bend, B2) and emerges subcutaneously in its third segment (S3) in the sub-occipital region. The deep cervical space course of the GON has never been assessed with MRI tractography. Therefore, the objective of this study was to assess the feasibility of GON MRI diffusion tractography at 3 T, in comparison with the anatomical atlas.
http://dx.doi.org/10.1016/j.neurad.2017.09.001 0150-9861/© 2017 Elsevier Masson SAS. All rights reserved.
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
2
Fig. 1. Greater occipital nerve anatomical description. The image on the left is an anatomical drawing representing the course of the GON. It arises from the posterior ramus of the C2 root through the C1–C2 foramen (origin), and then presents three segments (S1, S2 and S3) and two bends (B1 an B2). Three areas of vulnerability have been described: at it’s origin, first bend and emergence.The image on the left shows a ROI based tractogram of the GON, accurately depicting the whole course of the GON in the deep cervical spaces. Note that the acquisition is limited from the origin to the emergence, S3 is not seen.
Material and methods Subjects and data acquisition The study protocol was approved by our institutional review board for a prospective study in healthy subjects (IRB 5891). Healthy subjects aged 18 years or over were consecutively included in this study if they had: • no history or symptoms of brain or psychiatric disorder and no medical treatment; • undergone DWI and post-processing TWI. Two radiologists managed the subjects who required MR scans. ® We performed the scan on a 3 T MR imaging Philips ACHIEVA 3.0 T TX with a 32 channel head coil. The MR scan included transverse, sagittal and coronal T2weighted spin-echo sequences. The parameters of the diffusion sequence were: b value 1000 s/mm2 ; 32 directions; acquired voxel size: 2 mm isotropic; field of view 220 mm; single-shot spin-echo sequence; scan duration: 15 . An additional pair of b = 0 images (one with reversed phase encoding) was acquired for estimation of the inhomogeneity field [11]. Pre-processing and tractography reconstructions Diffusion image data (Fig. 2A) were corrected for B0 field in homogeneities and eddy current using top-up and eddy ® tools in FSL5 . The pre-processing steps were performed using MRtrix package software (JD Tournier, Brain Research Institute, Melbourne, Australia, http://www.brain.org.au/software/). Furthermore, for fiber orientation distribution estimation using the CSD model, the number of spherical harmonic terms was set to six. Whole-neck streamline tractography with 10 million streamlines
was then generated. Initial fractional anisotropy (FA) cut-off for terminating tracks was set to 0.3, an isotropic voxel size of 300 m and the minimum length of fibers to 20 mm as previously described in another cranial nerve tractography study [7]. To ensure that quantitative values yielded from DWI data were estimated in accordance with the SNR of the neck region, we excluded voxels with negative eigen-values before subsequent post-processing. TWI maps were performed (Fig. 2B) to produce an image of the mean length of fibers through each voxel rather than the count [12]. The post-processing lasted around 15 minutes using a com® puter with a multi-core processor (Intel Core i7, 3 GHz, Intel ® Corporation , USA). TWI maps analyses and GON study Two radiologists evaluated MRI tractography data including TWI and tractography reconstructions. The whole TWI tractography volume was used to assess GON visualization (Fig. 2B). For qualitative analyses, we visually evaluated the intensity of the GON with a three-grade ranking system. Each GON was assigned the following scores: • • • •
Score 0: no GON visualization; Score 1: Partial GON visualization; Score 2: Complete GON visualization; Score 2 was considered to represent physiological features. Scores 0 and 1 (Fig. 2) were considered as TWI failure.
The results of MR imaging studies of each patient were considered positive or negative for GON visualization by each observer. For an MR imaging–positive result, the GON had to be assigned a score of 2. Moreover, the following criteria were also assessed: position of B1 compared to the C2 pedicle (below or above lower aspect of
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
3
Fig. 2. Steps of greater occipital nerve tractography process with MRtrix. A. Raw DWI image at the C1–C2 foramen level after susceptibility and movement artifacts correction. B. Example of TWI mapping at the same C1–C2 level. This map was obtained after a first tractography process with generation of 10 millions fibers on the whole neck volume, then automatically converted into this picture file. High signal intensity (red) is correlated to the mean length of the tracted fiber. The direct visualization of GONs (white arrow heads) allow to ease extraction of coordinates before placing ROI. C. Bilateral GON (white arrow heads) tractogram after ROI extraction superimposed on morphological axial T2 weighted Imaging. D. Selective TWI GON representation (white arrow heads), with axial T2 superimposition. Finally, the track files were converted into GON picture files, with super-resolution technique, which allows GON representation with an isotropic resolution of 0.3 mm. This technique is useful for exploring the GON course in 3D slice by slice.
pedicle); intra- or extra-muscular course of B1 (through the inferior oblique muscle); visualization of motor branches; anatomical variations of the S2 segment of the GON. Inter-rater agreement on detecting GON was estimated using Cohen’s kappa coefficient. In addition, FA of the GON on both sides was assessed, using FA maps as a template as well as two regions of interest to extract the GON from TWI maps (Fig. 2C). A Student’s t-test was performed to assess the differences in FA between the two GON among subjects. Continuous statistical data were analyzed using SPSS software v22.0 (IBM, Inc., Armonk, New York, USA).
whether the course of B1 was situated intra- or extra-muscularly to the inferior oblique muscle in all cases. Motor branches were identified in only four cases. GON S2 segment assessment showed no variations in all 18 cases. Quantitative method by FA measurement GON fractional anisotropy was estimated to 0.47 ± 0.08 in subjects. There was no significant difference in FA measurements between the two GON in each diffusion acquisition. Discussion
Results Anatomical and clinical considerations GON visualization A total of 20 subjects were included (12 males and 8 females, mean age 53.8 yr). In comparison with the anatomical atlas, GON complete visualization (score 2) was possible in 18 out of 20 subjects. In two cases, GON was not visible in the cervical spine foramen (score 1). No subjects referred with score 0. Inter-rater agreement on detecting GON was estimated as being 1.0 on 2D reconstructions, implying that both radiologists were able to systematically detect GONs among the 18 subjects (kappa test), without discordance for the two remaining cases. In all but four cases, the first bend (B1) was depicted above the inferior aspect of the ipsilateral pedicle. It was not possible to assess
The anatomical course of the GON has been widely studied due to its implication in various types of headache syndromes [8,9]. The GON has been described to have three segments and two bends [13] (Fig. 1) with several possible areas of GON vulnerability [14]: first, at its origin from the C2 dorsal ramus between the atlas and the axis; second, at the first bend where the GON curves around the inferior oblique capitis muscle; third, at its superficial emergence when perforating the aponeurosis of the trapezius muscle. Anatomical studies of the GON have mainly focused on assessing the superficial emergence of the GON through the trapezius muscle, because this vulnerability site is the most used target site for GON infiltration in clinical practice [15–17] based on anatomical
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5 4
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
landmarks. The anatomical variability of the GON seems to be greater after the second bend and throughout the third segment of the GON [13,14,18]. However, only a few anatomical studies have assessed the initial segments of the GON and its variability [19]. Our results show that the GON is correctly and accurately depicted with TWI tractography. Moreover, our study showed the existence of subtle anatomical variations of the GON course in its initial bend in comparison with the C2 pedicle, however, no variations were found in the second segment of the GON (S2), during its course between the inferior oblique and capitis muscles. Previous studies have targeted the initial course of the GON (S1, B2 and S2) with satisfactory results [20,21]. Therefore, the presence of anatomical variations may be of interest to the interventional physician targeting the GON in some procedures, such as GON neurolysis, which has also been described [14,22]. In these specific situations, TWI tractography may appear as a useful pre-planning tool allowing detection of both possible anatomical variations, and accurate nerve localization as neurolysis procedures require immediate proximity between the needle and the nerve [23] for a successful outcome. In two cases, the GON was not seen in the C1–C2 foramina. DWI is prone to distortion and motion artifacts, which may be problematic for the study of GON in the foramina. However, to limit this problem we associated a method of diffusion optimization specifically adapted to correct for patient movements with global track-weighted imaging. GON motor branches were only detected in four of 18 cases. This might be explained by anatomical variability [14] of the motor branches. There are also technical limitations for depicting small nerves because the CSD model relies on probabilistic algorithms thatgive priority to large nerve trunk detection over small branches. We can also mention that CSD has proved to be valuable in overcoming issues of intraparotid VIIn variability and complex branching patterns. Moreover, CSD is based on orientation density fiber (ODF) reconstructions, known to be superior to diffusion ODF methods [19], in terms of tractography accuracy. Track-weighted imaging, a new qualitative tool to assess nervous pathways The TWI technique was recently introduced as a qualitative tractography method due to its high anatomical contrast [6]. In this acquisition, the nervous pathways can be combined with an anatomical reference map to facilitate nerve identification [7]. Whilst seed-based tractography is generally user-dependent, requiring in-depth structural neuroanatomical knowledge, a whole cervical fiber tracking approach offers a user-independent method for extracting visual information from the tractogram itself. The interest of FA measures in comparison with direct visual assessment of GON may be discussed. Quantitative FA analysis is currently the gold standard in tractography studies; TWI map does not yet appear as an ideal tool to provide quantitative values [24]. In addition, this study demonstrated no significant difference between GON FA values on both sides. It may be interesting to extract FA in patients with neuralgia, looking for FA decrease on the affected nerve in comparison with the contralateral side. Yet, the use of regions of interest (ROI) required by the FA method may be difficult in clinical practice and user-dependent, as ROI placement requires precise anatomical knowledge of GON. With this regard, it seems interesting to highlight the absence of GON discontinuity in 18 out of 20 subjects. Calamante et al. have previously validated TWI maps using direct comparison with mouse brain histological patterns [12]. In this work, the bright areas on the TWI maps significantly matched the dark areas in the myelin stained images. In another TWI study in patients with glioblastoma, higher relative TWI values of the tumor location yielded a higher likelihood of increased tumor proliferation, greater architectural
disruption, and microvascular hyperplasia [25]. Previous studies have shown the feasibility of such a technique on retinal tractography in healthy subjects [26]. In this GON study, we were unable to depict differences in the signal intensity of the nerve in the different subjects. However, future studies may depict modification of the GON in pathological conditions such as greater occipital neuralgia or other cranio-facial pain syndromes. Finally, and concerning the feasibility of TWI mapping in clinical practice, the acquisition time does not represent an issue, as the sequence used in this study lasts less than 10 minutes. Therefore, the main challenge is the possible movement artifact that may occur in some patients, although most artifacts can be dealt with in the post-processing steps. Post-processing is also a challenge in the everyday work flow, as 20 minutes are needed to edit the TWI maps. Limitations of this study Limitations of the study include the relatively small number of subjects included. Moreover, previous high-order tractography studies recommended the acquisition of more than 45 DW directions to avoid uniformity problems of diffusion weighted gradient directions, and to meet SNR requirements. However, this study showed successful identification of the GON in 18 out of 20 subjects. The method used is also user-dependent for FA extraction since it required anatomical knowledge concerning the GON course. However, the “Whole-Neck” streamlines approach with generation of tractograms in the neck area allowed an efficient identification of the GON course. In addition, the effect of patient age on FA measurement could be further assessed. Conclusion MRI diffusion tractography of the GON is feasible, allowing accurate depiction of the GON and may be applied in clinical practice. The usefulness of GON tractography in clinical routine is yet to be demonstrated. However, it may be very interesting in patients suffering from GON-mediated cranio-facial pain, in order to detect possible GON impairment. It may also appear as a potential preplanning tool in specific procedures targeting the GON. GON MRI diffusion tractography will need further investigations. Disclosure of interest The authors declare that they have no competing interest. References [1] Takagi T, Nakamura M, Yamada M, Hikishima K, Momoshima S, Fujiyoshi K, et al. Visualization of peripheral nerve degeneration and regeneration: monitoring with diffusion tensor tractography. Neuroimage 2009;44:884–92. [2] Kasprian G, Amann G, Panotopoulos J, Schmidt M, Dominkus M, Trattnig S, et al. Peripheral nerve tractography in soft tissue tumors: a preliminary 3-tesla diffusion tensor magnetic resonance imaging study. Muscle Nerve 2015;51:338–45. [3] Rouchy RC, Attyé A, Troprès I, Medici M, Kastler A, Righini C, Krainik A. Facial nerve tractography: a new tool to detect perineural invasion in parotid cancers. 10.1016/j.neurad.2017.01.005. [4] Farquharson S, Tournier JD, Calamante F, Fabinyi G, Schneider-Kolsky M, Jackson GD, et al. White matter fiber tractography: why we need to move beyond DTI. J Neurosurg 2013;118:1367–77. [5] Tournier JD, Calamante F, Gadian DG, Connelly A. Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution. Neuroimage 2004;23:1176–85. [6] Calamante F, Tournier JD, Kurniawan ND, Yang Z, Gyengesi E, Galloway GJ, et al. Super-resolution track-density imaging studies of mouse brain: comparison to histology. Neuroimage 2012;59:286–96. [7] Attye A, Karkas A, Tropres I, Roustit M, Kastler A, Bettega G, et al. Parotid gland tumours: MR tractography to assess contact with the facial nerve. European radiology 2015, http://dx.doi.org/10.1007/s00330-015-4049-9. [8] Anthony M. Headache and the greater occipital nerve. Clin Neurol Neurosurg 1992;94:297–301.
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001
G Model NEURAD-668; No. of Pages 5
ARTICLE IN PRESS A. Kastler et al. / Journal of Neuroradiology xxx (2017) xxx–xxx
[9] Blumenfeld A, Ashkenazi A, Napchan U, Bender SD, Klein BC, Berliner R, et al. Expert consensus recommendations for the performance of peripheral nerve blocks for headaches-a narrative review. Headache 2013;53:437–46. [10] Loukas M, El-Sedfy A, Tubbs RS, Louis Jr RG, Wartmann CH, Curry B, et al. Identification of greater occipital nerve landmarks for the treatment of occipital neuralgia. Folia Morphol (Warsz) 2006;65:337–42. [11] Andersson JL, Skare S, Ashburner J. How to correct susceptibility distortions in spin-echo echo-planar images: application to diffusion tensor imaging. Neuroimage 2003;20:870–88. [12] Pannek K, Mathias JL, Bigler ED, Brown G, Taylor JD, Rose SE. The average pathlength map: a diffusion MRI tractography-derived index for studying brain pathology. Neuroimage 2011;55:133–41. [13] Vital JM, Grenier F, Dautheribes M, Baspeyre H, Lavignolle B, Senegas J. An anatomic and dynamic study of the greater occipital nerve (n. of Arnold). Applications to the treatment of Arnold’s neuralgia. Surg Radiol Anat 1989;11:205–10. [14] Bogduk N. The anatomy of occipital neuralgia. Clin Exp Neurol 1981;17:167–84. [15] Dach F, Eckeli AL, Ferreira Kdos S, Speciali JG. Nerve block for the treatment of headaches and cranial neuralgias - a practical approach. Headache 2015;55(1):59–71. [16] Ambrosini A, Vandenheede M, Rossi P, Aloj F, Sauli E, Pierelli F, et al. Suboccipital injection with a mixture of rapid- and long-acting steroids in cluster headache: a double-blind placebo-controlled study. Pain 2005;118:92–6. [17] Weibelt S, Andress-Rothrock D, King W, Rothrock J. Suboccipital nerve blocks for suppression of chronic migraine: safety, efficacy, and predictors of outcome. Headache 2010;50:1041–4.
5
[18] Bovim G, Bonamico L, Fredriksen TA, Lindboe CF, Stolt-Nielsen A, Sjaastad O. Topographic variations in the peripheral course of the greater occipital nerve. Autopsy study with clinical correlations. Spine 1991;16:475–8. [19] Tubbs RS, Watanabe K, Loukas M, Cohen-Gadol AA. The intramuscular course of the greater occipital nerve: novel findings with potential implications for operative interventions and occipital neuralgia. Surg Neurol Int 2014;5:155. [20] Pougnard-Bellec F, Rolland Y, Morel D, Meadeb J, Marin F, Duvauferrier R. [Efficacy of C1-C2 block with posterior parasagittal approach in the treatment of Arnold neuralgia in 24 patients]. J Radiol 2002;83:133–9. [21] Aubry S, Kastler B, Bier V, Hadjidekov V, Hussein HH, Fergane B. Evaluation of the effectiveness of CT-guided infiltration in the treatment of Arnold’s neuralgia. Neuroradiology 2009;51:163–8. [22] Cohen SP, Peterlin BL, Fulton L, Neely ET, Kurihara C, Gupta A, et al. Randomized, double-blind, comparative-effectiveness study comparing pulsed radiofrequency to steroid injections for occipital neuralgia or migraine with occipital nerve tenderness. Pain 2015;156:2585–94. [23] Bogduk N, Dreyfuss P, Govind J. A narrative review of lumbar medial branch neurotomy for the treatment of back pain. Pain Med 2009;10:1035–45. [24] Jones DK, Knosche TR, Turner R. White matter integrity, fiber count, and other fallacies: the do’s and don’ts of diffusion MRI. Neuroimage 2013;73:239–54. [25] Barajas Jr RF, Hess CP, Phillips JJ, Von Morze CJ, Yu JP, Chang SM, et al. Superresolution track density imaging of glioblastoma: histopathologic correlation. AJNR Am J Neuroradiol 2013;34:1319–25. [26] Jean C, Attyé A, Grand S, Lecler A, Chiquet C, Aptel F, Lamalle L, Krainik A. Retinal tractography: influence of diffusion acquisition, ocular laterality and ocular dominance in healthy subjects. 10.1016/j.neurad.2017.01.035.
Please cite this article in press as: Kastler A, et al. Greater occipital nerve MR tractography: Feasibility and anatomical considerations. J Neuroradiol (2017), http://dx.doi.org/10.1016/j.neurad.2017.09.001