HE relationships of the cranial nerves and the vessels at the surface of the brainstem constitute a very com- plex 3D structure. 11–13,24,32 Hyperactive cranial nerve dysfunction syndromes resulting from neurovascular com- pression, such as trigeminal neuralgia, hemifacial spasm, and glossopharyngeal neuralgia as well as the association of hypertension with neurovascular compression are some en- tities that need a comprehensive 3D analysis and spatial un- derstanding of these relationships. Exact information about these relationships can be helpful for both diagnosis and preoperative surgical planning. Until now, imaging of the neurovascular structures in the posterior fossa and, especial- ly, of the relationship between cranial nerves and vessels has depended on 2D representations (slice images) of tomo- graphic volumes. 3,4,7 As an exclusive source of information the 2D slice images require extensive experience on the part of the observer to achieve a correct assessment, which may be incomprehensible to other observers. In many cases, this modality does not give a satisfactory overview of the un- derlying complex anatomy in the posterior fossa. Since the introduction of MR imaging, neurovascular re- lationships have been examined using several modalities including T 2 - and T 1 -weighted sequences, MR angiogra- phy, combinations of different imaging sequences within a protocol, and different orientations. 3,4,7,10,18,20,26,28,31,40,44,45,47 Im- provements in MR technology and progress in developing new MR imaging protocols like CISS resulting in strong- ly T 2 -weighted images 9 provide sufficiently high resolu- tion below 1 mm, even between consecutive slices. 23,27,30,41,48 The imaging data demonstrate improved contrast between neighboring structures, allowing us to differentiate crani- al nerves and vessels within the CSF at the surface of the brainstem. 9 Although MR CISS, fast turbo spin echo, and time-of-flight MR angiography produce 3D imaging data, it is important to point out the principal difference between 3D imaging data and 3D visualization. A number of stud- ies apply 3D MR sequences and their authors claim to have achieved 3D illustrations or visualizations. Nonetheless, this is misleading 7,10,20,28,44 in terms of computer graphics, be- cause 3D visualization comprises all methods of volume rendering, which allow us to extract and show 3D represen- tations of 3D imaging data (for example, MR imaging vol- umes). 1,6,14,15,42,43 The resulting illustrations should provide 3D objects that are close to reality based on the underlying imaging data. 2,5,14,16,36,43 Attempts at 3D visualization have been made in two studies, but neither of them presented any details about the applied principles and methods of image J. Neurosurg. / Volume 100 / June, 2004 J Neurosurg 100:1025–1035, 2004 Three-dimensional visualization of neurovascular relationships in the posterior fossa: technique and clinical application RAMIN NARAGHI, M.D., PETER HASTREITER, DR. ING., BERND TOMANDL, M.D., AGATHA BONK, M.S., W ALTER HUK, M.D., AND RUDOLF F AHLBUSCH, M.D. Department of Neurological Surgery, Neurocenter and the Division of Neuroradiology, University of Erlangen-Nuremberg, Germany Object. The goal of this study was to describe the authors’ technique for three-dimensional (3D) visualization of neurovascular relationships in the posterior fossa at the surface of the brainstem. This technique is based on the pro- cessing of high-resolution magnetic resonance (MR) imaging data. The principles and technical details involved in the accurate simultaneous visualization of vessels and cranial nerves as tiny structures are presented using explicit and im- plicit segmentation as well as volume rendering. Methods. In this approach 3D MR constructive interference in steady state imaging data served as the source for im- age processing, which was performed using the Linux-based software tools SegMed for segmentation and Qvis for vol- ume rendering. A sequence of filtering operations (including noise reduction and closing) and other software tools such as volume growing are used for a semiautomatic coarse segmentation. The subsequent 3D visualization in which im- plicit segmentation is used for the differentiation of cranial nerves, vessels, and brainstem is achieved by allocating opacity and color values and adjusting the related transfer functions. This method was applied to the presurgical eval- uation in a consecutive series of 55 patients with neurovascular compression syndromes and the results were correlat- ed to surgical findings. The potential for its use, further developments, and remaining problems are discussed. Conclusions. This method provides an excellent intraoperative real-time virtual view of difficult anatomical rela- tionships. KEY WORDS  neurovascular compression  cranial nerve  brainstem  segmentation  direct volume rendering T 1025 Abbreviations used in this paper: AICA = anterior inferior cer- ebellar artery; CISS = constructive interference in steady state; CSF = cerebrospinal fluid; MR = magnetic resonance; MVD = mi- crovascular decompression; PC = personal computer; PICA = poste- rior inferior cerebellar artery; RExZ = root exit zone; REZ = root en- try zone; 2D = two-dimensional; 3D = three-dimensional.