The Use of Intraoperative Navigation for Complex Upper Cervical Spine Surgery: Report of 4 Cases

Kern H. Guppy, M.D., Ph.D; Indro Chakrabarti, M.D., M.P.H; Ami t Banerjee, M.D.


Neurosurg Focus. 2014;36(3):e5 

In This Article


There is no doubt that intraoperative spinal imaging and navigation has been advanced with the introduction of the O-arm surgical imaging system and StealthStation navigation system.[11] The technology encompasses fluoroscopically acquired 2D images with a full 360° scan that is used for 3D volumetric reconstruction. An integrated LED (light-emitting diode) tracking system that incorporates the O-arm with the reference arc placed on the patient allows for automated registration as the images are acquired and provides the navigation component for the surgeon.[7]

The use of the O-arm with StealthStation guidance has been shown to be invaluable in increasing the accuracy of pedicle screw placement in the lumbar and sacral spine when compared with freehand placement,[21] as well as in pedicle screw placement in the thoracic and lumbosacral spine when compared with fluoroscopy-guided placement.[20,22] More recently it was used in the placement of distal pelvic fixation with bilateral S-2 alar iliac fixation.[16] It has also been shown to reduce operative time, in comparison with the use of preoperative CT scans with image guidance,[2] and to reduce radiation exposure compared with fluoroscopic guidance.[14] Performing intraoperative CT at the end of the surgical procedure also reduces the need to return patients to the operating room.[23] The O-arm has also been used in minimally invasive spine surgery[1,3] including endoscopic surgery for resection of chordomas[8] and intraoperative CT–navigated lateral interbody fusion procedures.[4] Its role in the cervical spine, however, has been limited, both for anterior cervical surgery and posterior cervical fusion.[9,15,18] The 4 cases presented in this paper illustrate new applications in the use of the O-arm for image guidance in the upper cervical spine.

Posterior Upper Cervical Spine Surgery Using Image Guidance

Because of the intimate spatial relationship between the vertebral artery and spinal cord in the upper cervical region, precise placement of screws is essential. Nottmeier and Young[15] described a technique for placing screws at the occiput, C-1, and C-2 using 2 paired image guidance systems: 1) the BrainLAB VectorVision system in conjunction with the Arcadis Orbic Isocentric C-arm (Siemens Medical Solutions), and 2) the O-arm paired with the Stealth Treon system (Medtronic). They reported on placement of a total of 82 screws, including 24 in the occiput, 24 in lateral masses of C-1, 13 in the C-2 pars, and 21 in C-2 laminae, with only 1 screw with a minimal breach of the outer lamina of C-2.

Two years later, Yu et al.[25] described 108 screw insertions performed for the treatment of complex craniovertebral junction malformations using a CT scanner installed in a preexisting operating room connected to an integrated neuronavigation system. There were 20 C-1 lateral mass screws, 26 C-2 screws, 14 C-3 or C-4 pedicle screws, and 32 occipital screws placed. Their accuracy, using image guidance, was 98.1%.

Ishikawa et al.[9] retrospectively reviewed 21 consecutive cases in which 108 cervical pedicle screws were placed using the O-arm-based navigation system. They reported a minor pedicle violation rate of 8.3% (9 of 108 screws) and a major pedicle violation rate of 2.8% (3 of 108 screws). While these error rates may be low, we believe that if the misplacements had been found intraoperatively they could have been corrected. For this reason we have advocated performing another intraoperative CT scan before wound closure. Van de Kelft et al.[23] repositioned approximately 1.8% of their screws that had been placed with CT-guided navigation.

In our series we used the O-arm to place 6 occiput screws (Cases 1 and 2), 5 posterior C-1 lateral mass screws (Cases 2, 3, and 4), 2 C-2 translaminar screws (Case 4), 2 C-6 pedicle screws (Case 3), 2 C-7 pedicle screws (Case 1), and 2 anterior lateral mass screws at C-1 (Case 3). Repeat intraoperative CT before wound closure showed no pedicle violations or malposition of any of the 19 screws we placed using the O-arm system.

Odontoidectomy Using Image Guidance

Veres et al.[24] in 2001 described their early use of navigation-assisted transoral odontoidectomy in 3 patients. Preoperatively the patients were placed in a halo device, with fiducial markers attached supraorbitally and to both mastoids, and a CT scan was obtained. Resection of the odontoid was accomplished using the BrainLAB Vector-Vision navigation system with little need for fluoroscopy. More recently, in 2012, Li et al.[10] described the use of a ceiling-mounted frameless infrared-based neuronavigation system (VectorVision) and 40-multislice CT scanner mounted in an operating room in 19 cases involving patients with complex craniovertebral junction malformations. Eighteen of these patients had a transoral approach with image guidance combined with endoscopy.

There have been no reported cases in which the O-arm for was used for odontoidectomies. In this article, we present, for the first time, the use of CT image guidance (O-arm), as shown in Case 2, by defining the superior and inferior margins as well as the depth for complete removal of the odontoid. We have also found that a second intraoperative CT scan is necessary to evaluate the bony decompression. This is far superior to the technique described by Mummaneni and Haid[12] in which iohexol dye was placed into the resection cavity after an odontoidectomy and a lateral fluoroscopic image was obtained to confirm the extent of decompression. The authors found that the spread of the dye helped to reveal any remaining remnant of the dens. Li et al.[10] also described the usefulness of a second intraoperative CT study in their Patient 5 in whom a postdecompression CT was performed and a small residual bone spicule compressing the neuraxis was found. The spicule was removed, and a third intraoperative 3D-CT showed that sufficient anterior decompression had been achieved.

Chordoma Resection Using Image Guidance

Chordoma resections in the cervical spine can be difficult due to the predilections of these tumors to occur near delicate and vital structures and their destructive behavior, which can obscure familiar bony landmarks. Hsu et al.[8] described the use of an image-guided, endoscopic, transcervical approach for resection of a recurrent chordoma at C1–2. They used the BrainLAB frameless stereotactic system to remove bone, tumor, and methylmethacrylate from a previous reconstruction. They did not place any spinal instrumentation using image guidance. En bloc resection of chordomas with negative margins can be more difficult to perform without damaging adjacent structures and causing significant clinical morbidity. We previously described the technique of en bloc resection of an upper cervical chordoma using image guidance.[6] The tumor distorted the normal anatomy, and CT guidance allowed en bloc resection without violating vital structures. In Case 3, we further describe details of the use of the O-arm–based system for placement of the occipital screws, lateral mass screws at C-1, and pedicle, as well as resection of the chordoma using image guidance.

Revision Cervical Spine Surgery Using Image Guidance

In situations where anatomical landmarks are difficult to identify, CT image guidance has been shown to be ideally suited. In patients with scoliosis, degenerative spine disease, or ankylosing spondylitis and in very obese patients or in treating areas of the cervicothoracic junction and upper thoracic spine where fluoroscopy has poor resolution, CT guidance is very useful with reduction in radiation exposure.[5,18] This is especially true in revision cervical spine surgery where the normal anatomical landmarks are obscured. Seichi et al.[19] used a frameless stereotactic image-guidance system with simultaneous fluoroscopy in 10 patients who required revision of previous cervical spine surgery. They found image guidance helped them to achieve rigid fixation and ensure safety. We also found it useful as illustrated in Case 4. The revision of the C1–2 nonunion in that case was challenging due to bony overgrowth that obscured the normal landmarks for the placement of the C-2 laminar screws. In Case 1 also, bony landmarks were not apparent, but this time it was due to a congenital deformity. In that case, the deformity in the lateral mass distorted the normal anatomy, and intraoperative CT was essential in locating the lateral mass and removing it.

Technical Nuances

Since we acquired the O-arm surgical imaging system in 2006, several technical nuances have helped us to reduce the time to obtain imaging, to maintain a sterile environment, and to increase accuracy.

To reduce the time to obtain and transmit images, we have found it important to train our staff, as well as the surgeons, to become familiar with the equipment and to be able to troubleshoot error messages. As the radiology technicians became better versed with the use of the O-arm system, performing a CT scan and transferring images to the Medtronic StealthStation navigation system could be accomplished in 20 minutes.

Maintaining sterile conditions was also very important. Using a plastic cover for the O-arm "donut" (Fig. 11B and C) provided such an environment. We have found that covering the operative field with sterile drape sheets and then bringing the O-arm "donut" uncovered was faster and still maintained a sterile field.

Figure 11.

O-arm and patient positioning. A: Patient in Mayfield head clamp with support/mounting device (white arrow) reversed and with Mayfield Tri-Star swivel adaptor (black arrow). B and C: Patient in low profile to fit into the "donut" of the O-arm, with reference arc (black arrow) attached to Tri-Star swivel adaptor.

Accuracy was achieved by proper placement of the reference arc as well as minimizing movement of the vertebrae. The placement of the reference arc increased our accuracy in screw placement if it was placed close to the structure of interest (Fig. 11). As Nottmeier and Pirris[13] pointed out in their article, placement of the reference arc attached to the Mayfield frame provided accuracy in screw placement for the occiput to C-3 and C-4. The further away from the reference arc, the less accurate the placement was. A second intraoperative CT scan may need to be obtained with the arc placed on a cervical spinous process. In our series only one intraoperative CT scan was needed for placement of screws in all cases but Case 2; in Case 2, pedicle screws were placed in the upper thoracic spine, requiring the arc to be placed on C-7 for a second intraoperative CT. Also, although in several articles[13] the Mayfield frame was placed on the prone patient and attached to the Jackson table, we found it works just as well on the typical neurosurgery operating table (for example, Skytron) as long as the Mayfield support/mounting device is reversed so as to have a low profile to fit into the donut of the O-arm (Fig. 11A). Removal of the tabletop cushion above the thoracic chest, as well as using small thoracic bolsters, reduces the height.

Limiting intersegmental movement of the vertebrae by placing the retractors prior to the intraoperative CT was also very important. Of note, the C-1 vertebra is quite mobile, and C-1 screws should not be advanced while following the navigation. The trajectory should be used and the screws advanced in small steps; navigation should be used when the small screw advancement is temporarily stopped. We have found this to be a problem mainly at C-1.

Finally, we must remember that CT image guidance should be used as an aid in spine surgery. It does not replace anatomical knowledge, which should always be used, if possible, when placing hardware. We have also found that frequent checking of familiar landmarks keeps the surgeon aware of any changes in the accuracy of the guidance system.