Journal of Neuroimaging Vol 11 No 2 April 2001Seitz et al: IXth to XIIth Cranial Nerves in MRI

Visualization of the IXth to XIIthCranial Nerves Using3-Dimensional ConstructiveInterference in Steady State,3-DimensionalMagnetization-Prepared RapidGradient Echo and T2-Weighted2-Dimensional Turbo SpinEcho Magnetic ResonanceImaging Sequences

Johannes Seitz, MD

Paul Held, MD

Rüdiger Fründ, PhD

Michael Strotzer, MD

Wolfgang R. Nitz, PhD

Markus Völk, MD

Thomas Haffke

Stefan Feuerbach, MD

A B S T R A C T

Objective. The purpose of this study was to evaluate the visibilityof the IXth to XIIth cranial nerves using different magnetic reso-nance sequences. Thirty healthy volunteers underwent mag-netic resonance imaging at 1.5 T using 3-dimensional construc-tive interference in steady state (CISS) sequence (TR = 17 ms,TE = 8.08 ms, α = 70°), 3-dimensional magnetization-preparedrapid gradient echo (MP-RAGE) sequence (TR = 11.08 ms, TE =4.3 ms, α = 15°), and T2-weighted (w) 2-dimensional turbo spinecho (TSE) sequence (TR = 4000 ms, TE = 102 ms, α = 180°,slice thickness = 2 mm). Visibility of the IXth to XIIth cranialnerves in each sequence was evaluated by consensus of 2 radi-ologists using an evaluation scale from 1 (excellently visible) to 5(not visible). A correlation with anatomic specimens was made.The 3-dimensional CISS sequence provides best resolution ofthe IXth to XIIth cranial nerves and their relation to surroundingstructures. Additional information is given by the 3-dimensionalMP-RAGE when nerves are surrounded by soft tissues. Usingthe T2w 2-dimensional TSE sequence, even whole nerves can-not be visualized due to intersection gap and partial volumeeffects. However, even in 3-dimensional high-resolutionsequences, segments of nerves are not always visualized. A

combination of 3-dimensional CISS and 3-dimensionalMP-RAGE proved to be useful to visualize the IXth to XIIth cranialnerves, whereas the 2-dimensional technique failed. Furtherinvestigations using 3-dimensional MP-RAGE with contrastmedium should be performed in the case of abnormality.

Key words: Magnetic resonance imaging (MRI), glossopha-ryngeal nerve, vagus nerves, spinal accessory nerve,hypoglossal nerve, anatomy.

Seitz J, Held P, Fründ R, Strotzer M, Nitz WR,Völk M, Haffke T, Feuerbach S.

Visualization of the IXth to XIIth cranial nerves using3-dimensional constructive interference in steady

state, 3-dimensional magnetization-preparedrapid gradient echo and T2-weighted 2-dimensional

turbo spin echo magnetic resonance imaging sequences.J Neuroimaging 2001;11:160–164.

In former times, the visualization of the cranial nervesrequired invasive techniques such as pneumo-encephalography or high-resolution tomodensitometrywith negative contrast (air) or positive contrast (nonioniccontrast agents). Since magnetic resonance imaging(MRI) was developed, some attempts have been made tostudy the larger cranial nerves (eg, optic nerve II,trigeminal nerve V, facial nerve and cochlear nerve VII &VIII) and their pathologies.1,2 In recent years, sequencesproviding high spatial resolution in combination with agood signal-to-noise ratio were evaluated allowing thedetection of structures smaller than 1 mm in size.

The aim of this study was to evaluate the anatomy ofthe intracranial course of the glossopharyngeal (IX),vagus (X), spinal accessory (XI), and hypoglossal (XII)

160 Copyright © 2001 by the American Society of Neuroimaging

Received May 26, 2000, and in revised form July 19,2000. Accepted for publication September 26, 2000.

From the Department of Diagnostic Radiology, Univer-sity Hospital of Regensburg, Germany ( JS, PH, RF, MS,WRN, MV, SF); and the Institute of Anatomy, Universityof Regensburg, Germany (TH).

Address correspondence to Dr Seitz, Department of Diag-nostic Radiology, University Hospital , 93042Regensburg, Germany. E-mail: johannes.seitz@klinik.uni-regensburg.de.

nerves using T1-weighted (w) and T2*w 3-dimensionaland transverse T2w 2-dimensional sequences.

Materials and Method

Thirty healthy volunteers (15 women and 15 men) with amean age of 32 years (from 18 to 53 years) underwentMRI using a 1.5-T scanner (Magnetom Symphony,Siemens, Erlangen, Germany) with 20 mT/m gradientstrength and a circularly polarized head coil. Thus, 60glossopharyngeal, vagus, spinal accessory, and hypo-glossal nerves were studied. Informed consent wasobtained. A protocol consisting of the followingsequences was applied.

1. Sagittal T1w 3-dimensional magnetization-preparedrapid gradient echo (MP-RAGE) (TR = 11.08 ms, TE =4.3 ms, TI = 300 ms, flip angle = 15°, bandwidth = 130Hz/pixel, effective slice thickness = 1 mm, pixel size =1.2 × 0.9 mm, acquisition time = 7 minutes, 22 seconds).This sequence is a refined version of a 3-dimensionalFLASH sequence structure,3 where the inversion pulse isplaced prior to each partition loop in order to introduceT1w. The effect of preparation is usually lost over thelength of a 3-dimensional acquisition. Therefore, 3-di-mensional MP-RAGE acquires all Fourier lines along thedepth-encoding loops, allows a recovery period, appliesthe preparation pulse again, and continues with the nextin-plane phase-encoding step.4

2. Paratransverse T2*w three-dimensional constructive in-terference in steady state (CISS) (TR = 17 ms, TE = 8.08ms, flip angle = 70°, bandwidth = 130 Hz/pixel, effectiveslice thickness = 1 mm, pixel size = 0.6 × 0.45 mm, acqui-sition time = 7 minutes, 51 seconds). Three-dimensionalCISS is based on a 3-dimensional gradient echo se-quence with gradient refocusing in all 3 directions (truefast imaging with steady-state free precession) generatinga steady-state contribution of the transverse magnetiza-tion for tissues with a long T2 relaxation time and is there-fore called T2w.5 To minimize the destructive interferencepattern of the originally published fast imaging withsteady-state free precession, 2 sequences can be executedand added together with an alternation of the phase ofthe radiofrequency pulse. Because the interference pat-terns are shifted due to the alternation, they almost van-ish when images are combined.4

3. Paratransverse T2w 2-dimensional turbo spin echo (TSE)(TR = 4000 ms, TE = 102 ms, flip angle = 180°, band-width = 130 Hz/pixel, effective slice thickness = 2 mm,intersection gap = 0.2 mm, number of acquisitions = 4,pixel size = 0.6 × 0.45 mm, acquisition time = 7 minutes,2 seconds). TSE sequences use multiple spin echoes withdifferent phase-encoding steps to fill the k-space fasterthan the single-echo approach in conventional spin echoimaging.4

For visualization of the IXth to XIIth cranial nerves,the 3-dimensional sequences were reconstructed in arbi-trary planes adapted to the course of each nerve using theimplemented evaluation software of the scanner. Thedetectability of the anatomy of these cranial nerves inboth reconstructed 3-dimensional sequences and in the

para-axial images of the 2-dimensional sequence wasevaluated by the consensus of 2 experienced radiologistsusing a 5-point scale (1 = excellently visible, 2 = well visible,3 = sufficiently visible, 4 = barely visible, 5 = not visible). Thefollowing structures were analyzed:

1. Glossopharyngeal nerve (at the exit and in its course inthe perimedullary cistern, and in the pars nervosa of thejugular foramen);

2. Vagus nerve (at the exit and in its course in theperimedullary cistern, and in the pars vascularis of thejugular foramen);

3. Spinal accessory nerve (medullary parts in theperimedullary cistern toward the pars vascularis of thejugular foramen, and spinal roots in the perimedullarycistern toward the pars vascularis of the jugular foramen);

4. Hypoglossal nerve (at the exit and in its course in theperimedullary cistern, and in the hypoglossal canal).

The 2-tailed Wilcoxon Test (confidence level 95%) wasused for statistical evaluation.

The nerves were identified in correlation with ana-tomic cadaver specimens especially made for this studyby Dr Herbert Hees (professor of anatomy at the Univer-sity of Regensburg). When in doubt, the nerve in questionwas discussed with the professor and identified in consen-sus direct on the evaluation console dynamically follow-ing the anatomic course.

Results

The mean values and standard deviations of the originand cisternal course toward the jugular foramen of theglossopharyngeal, vagus, spinal accessory, and hypo-glossal nerves examined with T2*w 3-dimensional CISS,T1w 3-dimensional MP-RAGE, and T2w 2-dimensionalTSE sequences are shown in Table 1. The best sequencefor the delineation of the IXth to XIIth cranial nervesin their cisternal course and their differentiation is 3-di-mensional T2*w CISS. The glossopharyngeal (IX) andvagus (X) nerves in most cases are excellently or well visi-ble at their origin, in the perimedullary cistern, and in thejugular foramen (Fig 1A). In the 3-dimensional T1wMP-RAGE sequence, the IXth and Xth cranial nervesgenerally are only barely visible or not visible at all(Fig 1B). In the 2-dimensional T2w TSE sequence, these 2nerves are often only sufficiently visible (Fig 1C) or barelyvisible in their intracranial course. They are missed in sev-eral cases.

In most cases, the spinal accessory (XI) andhypoglossal (XII) nerves are well or adequately visible in3-dimensional CISS images, with the exception of the spi-nal roots of the XIth cranial nerve, which are only barelyvisible. In the 2-dimensional TSE sequence, these 2 nerveswere judged as generally nonvisible with the exception of

Seitz et al: IXth to XIIth Cranial Nerves in MRI 161

the medullary parts of the spinal accessory nerve, whichare barely visible (Fig 2). In the 3-dimensional MP-RAGEtechnique, there is a poor visualization of the XIth andXIIth cranial nerves.

Discussion

The direct delineation of the IXth to XIIth cranial nervesis difficult because of their small dimension and close rela-tion from their origin and their common course in theperimedullary cistern toward the jugular foramina (Fig 3).To our knowledge, there is no study concerning the detec-tion of these nerves with different 3-dimensional and 2-di-mensional sequences and no prospective studyconcerning the visualization of these nerves.

Two high-resolution 3-dimensional MRI sequences(3-dimensional CISS and 3-dimensional MP-RAGE) andone 2-dimensional sequence (3-dimensional TSE) wereperformed, attempting to delineate the anatomy of eachlower cranial nerve.

Three-dimensional sequences were selected becauseof their relatively short acquisition time compared withthin section and 2-dimensional sequences in at least 3

main planes. A further advantage is the possibility of arbi-trary reconstructions and gapless image acquisition.1,2

The T1w MP-RAGE was selected because of its highsoft tissue contrast. The T2*w CISS has been widelyaccepted for the imaging of the facial (VII) andvestibulocochlear (VIII) nerves because of its high sig-nal-to-noise ratio and its high contrast of cerebrospinalfluid to nonliquid structures. Using these sequences, struc-tures with less than 1 mm in dimension, such as thecochlear, facial, superior, and inferior vestibular nerves inthe inner auditory canal, can be differentiated.1,2

The 2-dimensional T2w TSE sequence was chosenbecause it is a sequence commonly used in the examina-tion of the brain stem.

As a result of its high contrast, 3-dimensional CISS isthe best sequence for the visualization of the IXth to XIIthcranial nerves in their intracranial courses through theperimedullary cistern toward the jugular foramen. Theeffective slice thicknesses in the submillimeter range withnearly isotropic voxel sizes and the possibility of anatomi-cally adapted arbitrary reconstructions for each nerve arefurther advantages of both 3-dimensional sequences usedin this study.

162 Journal of Neuroimaging Vol 11 No 2 April 2001

Table 1. Detectability of Structures (mean values, standard deviations, and P values calculated with the Wilcoxon Test)

Two- Two-Dimensional CISS/MP- CISS/Two- Dimensional

MP-RAGE CISS TSE RAGE Dimensional TSE TSE

Glossopharyngeal nerveAt the exit and in its coursein the perimedullary cistern 3.13 ± 0.85 1.33 ± 0.63 3.17 ± 0.74 P < .0001 P < .0001 P = .786

In the pars nervosa of thejugular foramen 4.18 ± 0.97 1.83 ± 1.11 3.55 ± 0.93 P < .0001 P < .0001 P < .0001

Vagus nerveAt the exit and in its coursein the perimedullary cistern 3.58 ± 1.08 1.45 ± 0.79 3.25 ± 0.84 P < .0001 P < .0001 P = .036

In the pars vascularis of thejugular foramen 4.22 ± 0.98 2.00 ± 1.22 3.67 ± 1.02 P < .0001 P < .0001 P = .003

Spinal accessory nerveMedullary parts in theperimedullary cisterntoward the pars vascularisof the jugular foramen 3.68 ± 1.35 2.25 ± 1.46 4.02 ± 1.14 P < .0001 P < .0001 P = .089

Spinal roots in theperimedullary cisterntoward the pars vascularisof the jugular foramen 4.23 ± 0.95 3.57 ± 1.35 4.90 ± 0.35 P = .001 P < .0001 P < .0001

Hypoglossal nerveAt the exit and in its coursein the perimedullary cistern 3.80 ± 0.99 2.82 ± 1.52 4.60 ± 0.85 P = .001 P < .0001 P < .0001

In the hypoglossal canal 4.08 ± 1.03 2.98 ± 1.57 4.62 ± 0.85 P < .0001 P < .0001 P = .001

MP-RAGE = magnetization-prepared rapid gradient echo, CISS = constructive interference in steady state, TSE = turbo spin echo. The scale was 1 =excellently visible, 2 = well visible, 3 = sufficiently visible, 4 = barely visible, 5 = not visible.

Three-dimensional MP-RAGE is inferior to 3-dimensional CISS and 2-dimensional TSE in the visual-

ization of the IXth and Xth cranial nerves. This isexplained by lower fluid-to-soft tissue contrast compared

Seitz et al: IXth to XIIth Cranial Nerves in MRI 163

Fig 2. Elected picture of a T2w transverse 2-dimensionalturbo spin echo sequence (w = weighted): hypoglossal nerves(12) from their exits and their courses through the peri-medullarycistern (p)and in thehypoglossal canals (hc)arevisible.

Fig 1. (A) T2*w 3-dimensional constructive interference insteady-state sequence, paracoronal reconstructed (w = weighted):vagus (9), glossopharyngeal (10), and spinal accessory (11)nerves in their courses through the perimedullary cistern (p)toward the jugular foramen (j). (B) T1w 3-dimensional magneti-zation-prepared rapid gradient echo sequence, paracoronalreconstructed: vagus (9) and glossopharyngeal (10) nerves intheir course through the perimedullary cistern (p) toward thejugular foramen (j). (C) T2w transverse 2-dimensional turbospin echo sequence: glossopharyngeal nerves (10) andmedullary parts of the spinal accessory nerves (11) in theircourses from their exits through the perimedullary cistern (p)toward the jugular foramen (j).

Fig 3. Anatomic cadaver specimen (view from lateral dorsalcorresponding to the orientation of Figs 1A and 1B): vagus (9),glossopharyngeal (10), and spinal accessory (11) nerves intheir course through the perimedullary cistern (p) toward thejugular foramen; the hypoglossal (12), trigeminal (5), and co-chlear (8) nerves are visible.

to T2w and T2*w sequences. The vagus nerve is the largestof the cranial nerves studied in our investigation; thus, it isoften distinguishable even in 2 mm thick slices, and theadvantage of arbitrary thin reconstruction does not com-pensate the higher contrast.

In the detection of the XIth and XIIth cranial nerves,3-dimensional MP-RAGE is inferior to 3-dimensionalCISS but superior to 2-dimensional TSE. This resultsfrom the smaller perimedullary cistern in the lower por-tion near the foramen magnum and the closer correlationto soft tissue such as the spinal roots of the spinal accessorynerve coming up through the foramen magnum and thehypoglossal nerve in the hypoglossal foramen. Also, theintersection gap of the 2-dimensional TSE sequence leadsto a loss of information in these small structures. How-ever, even in 3-dimensional high-resolution sequencesand adapted reconstructions, segments of cranial nervesare not always visualized.

Conclusion

A combination of 3-dimensional CISS and 3-dimensionalMP-RAGE seems to be a useful protocol to study theintracranial anatomy of the IXth to XIIth cranial nerves.

References

1. Held P, Fellner C, Fellner F, Seitz J, Strutz J. MRI of innerear anatomy using 3D MP-RAGE and 3D CISS sequences.Br J Radiol 1997;70:465–472.

2. Casselman JW, Kuhweide R, Ampe W, Meeus L, SteyaertL. Pathology of the membranous labyrinth: comparison ofT1- and T2-weighted and gadolinium-enhanced spin-echoand 3DFT-CISS imaging. AJNR 1993;14:59–69.

3. Haase A. Snapshot FLASH MRI: applications to T1-, T2-and chemical shift imaging. Magn Reson Med 1990;13:77–89.

4. Nitz WR. MR imaging: acronyms and clinical applications.Eur Radiol 1999;9:979–997.

5. Deimling M, Laub GA. Constructive interference in steadystate for motion sensitivity reduction [abstract]. In: Book ofAbstracts. Vol 1. Berkeley, CA: Society of Magnetic Reso-nance in Medicine; 1989:642.

164 Journal of Neuroimaging Vol 11 No 2 April 2001