Virtual Fluoroscopy Overview and Future Implications

Y Raja Rampersauda, Kevin T. Foleyb c a Divisions of Orthopaedic and Neurosurgery, University of Toronto,

Toronto, Canada; b Division of Neurosurgery, University of Tennessee and c Image-Guided Surgery Center, Memphis, Tenn., USA

The utilization of neuronavigation in spinal surgery continues to grow. Numerous advances in this field have facilitated more practical clinical applications and have led to greater acceptance of this technology. The successful combination of neuronavigational technology with a practical intraoperative imaging modality (i.e. fluoroscopy) has significantly increased the clinical utility of computer-assisted surgery. Consequently, the introduction of computerassisted fluoroscopy ('virtual fluoroscopy') has had a significant impact across multiple surgical specialties.

The complex anatomy of the spine necessitates reliable preoperative and intraoperative imaging. In spinal surgery, a significant number of surgical complications are a direct result of poor intraoperative anatomical localization. However, due to technical limitations and cost considerations, detailed intraoperative imaging modalities such as computed tomography (CT) or magnetic resonance imaging (MRI) are not readily available. Consequently, the need to improve intraoperative visualization has led to the development of image-guided navigational systems for spinal surgery. The first such system was three-dimensional (3D) and relied on preoperative imaging (CT based) [1, 2]. Although conventional 3D image guidance technology has gained significant recognition and acceptance, it is by no means widely utilized by spine surgeons. This technology is still considered to be early in its evolution and is not without limitations. From a clinical perspective, the greatest limitation of current 3D systems is the time-consuming and often frustrating process of registration. Another limitation is the current inability to update the preoperative imaging to reflect the intraoperative position of the spine. At present, intersegmental changes in the position of the spine either due to a change in the patients' overall position compared to their preoperative position or following a reduction maneuver cannot be updated without the presence of an intraoperative CT or MRI. This limitation mandates that each segment be individually registered to provide maximal navigational accuracy at that respective segment. As previously stated, this can often be a rate-limiting step to the novice user of a conventional 3D image guidance system. Furthermore, the accuracy provided by these systems is often felt to be unnecessary for the majority of spinal procedures. These limitations continue to drive the need to develop more user-friendly and practical computer-assisted techniques, such as virtual fluoroscopy.

C-arm fluoroscopy is an intraoperative imaging technique that is familiar to all spine surgeons. It is routinely employed for real-time intraoperative localization of patient anatomy and surgical instruments in a variety of spinal procedures. Fluoroscopic localization facilitates improved accuracy and in many instances reduces surgical exposure in a variety of spinal procedures such as pedicle screw insertion, interbody cage placement, odontoid screw insertion, and atlantoaxial transarticular screw fixation. This imaging technology has also enabled the development of percutaneous spinal procedures, such as vertebro-plasty. Despite the advantages of intraoperative fluoroscopy, there are several limitations. Without a second fluoroscope, only a single projection can be visualized at one time. This limitation makes it necessary to reposition the C-arm during procedures that require multiple planes of visualization. Frequent repositioning of the C-arm is tedious, time-consuming, and frustrating. In addition, maintaining ideal sterility is a challenge. Furthermore, the position(s) of the C-arm is often ergonomically challenging to the surgeon. Finally, the potential deleterious effect of repeated radiation exposure to the surgeons' hands is always an important consideration [3].

Combining neuronavigational technology with a standard C-arm fluoroscope has enabled the optimization of this versatile intraoperative imaging modality. A computer-assisted image-guided fluoroscopy (virtual fluoroscopy) system provides the user with real-time multiplanar anatomical localization of a tracked surgical instrument in relation to patient-specific fluoroscopic-based images [4-8]. These systems eliminate the need for special preoperative imaging studies, manual image to patient registration (see above), C-arm repositioning, or the associated ergonomic challenges of using a C-arm fluoroscope. Furthermore, patient and occupational radiation exposure is significantly reduced, as is the need for wearing lead protection. Virtual fluoroscopy is currently being used for a wide variety of spinal applications. It is also being used for assistance in osseous cranial navigation (e.g. transsphenoidal access to the parasellar region) [9]. In addition, these systems are being extensively used for a variety of orthopedic trauma procedures and more recently total joint replacement. A novel variant of this technology is also being utilized in interventional radiology.

System Overview

A typical virtual fluoroscopy system consists of a surgical navigational computer system, a commercially available digital C-arm fluoroscope, a calibration target that attaches to the C-arm, and a variety of modified surgical instruments that are capable of being tracked by the image guidance system. The process of virtual fluoroscopy can be divided into four basic steps: (1) acquisition of one or more fluoroscopic images, (2) capturing the position of the patient and C-arm at the time of image acquisition, (3) mathematically recreating the fluoroscopic image formation process of the C-arm for each image acquired, and (4) superimposing the relative position of a tracked surgical instrument onto the virtual fluoroscopic image(s) [6, 8]. In step 1, conventional fluoroscopic images acquired in any plane are automatically transferred to the computer for image processing. In step 2, information about the relative position of the patient and the C-arm at the time of image acquisition (i.e. generation of the fluoroscopic image) is captured by a position-sensing camera array. The most commonly used tracking device is an electro-optical camera that detects the position of light-emitting diodes or passive reflectors attached to the object(s) of interest. Other means of position sensing (e.g. electromagnetic) can also be used. The patient position is defined by attachment of a dynamic reference array (DRA), which rigidly attaches to the portion of the patient's anatomy that is to be imaged. The C-arm position is defined by position markers or signal emitters built into the calibration target that is affixed to the image intensifier. In step 3, utilizing a mathematical algorithm, the computer recreates the fluoroscopic image formation process of the C-arm (i.e. virtual fluoroscopy). The algorithm calibrates the acquired fluoroscopic image by taking into account the positional data acquired in the second step. The calibration process compensates for factors such as gravity-dependent changes in the C-arm image center, the effect of external electromagnetic fields generated by electrical equipment in the operating room, and the effect of changes in the C-arm's position with respect to the earth's magnetic field. These factors are unique to every C-arm position; therefore, it is necessary that every acquired image be independently calibrated. Essentially the computer creates multiple live virtual flu-oroscopic beams that are maintained in their original spatial orientation relative to the patient. In step 4, the computer determines the position of one or more trackable surgical instruments using a position-measuring camera and then superimposes an image of the instrument(s) in the virtual fluoroscopic display. Dedicated image-guided instruments such as awls, probes, taps, and screwdrivers are commercially available. Alternatively, any rigid surgical instrument may be customized with a tracking device or tracked with the assistance of a universal tool array. The system can display the position of the surgical instrument(s)

Fig. 1. AP and lateral virtual fluoroscopic views of the lumbar spine. Note the highgrade spondylolisthesis. a The pedicle probe (dark line) has been positioned at the pedicle entry point (white arrowhead). Its trajectory (light line) has been virtually extended 15 mm to the base of the pedicle (black arrow). b Pedicle probe advancement is displayed in both the AP and lateral images simultaneously.

Fig. 1. AP and lateral virtual fluoroscopic views of the lumbar spine. Note the highgrade spondylolisthesis. a The pedicle probe (dark line) has been positioned at the pedicle entry point (white arrowhead). Its trajectory (light line) has been virtually extended 15 mm to the base of the pedicle (black arrow). b Pedicle probe advancement is displayed in both the AP and lateral images simultaneously.

in any of the previously acquired fluoroscopic images, in multiple planes, simultaneously. The system also allows the actual projection of a surgical instrument (in one color) and the simultaneous projection of the linear extension of that instrument's proposed trajectory (in a second color) (fig. 1).

Accuracy Validation

Bench top testing of these systems has shown excellent (submillimetric) correlation of the live fluoroscopic position of a probe and its virtual fluoroscopic projection. In a simulated operating room setting, Foley et al. [4] investigated the in vitro accuracy of the FluoroNav (Medtronic Surgical Navigation Systems, Louisville, Colo., USA) system. In this study, the difference in positioning of an implanted pedicle probe (tip and trajectory angle) was measured comparing live and virtual fluoroscopic images. The mean error in probe tip localization was 0.97 ± 0.40 mm (99% confidence interval = 2.2 mm, maximum probe tip error = 3 mm). The mean trajectory angle difference between the virtual and actual probe images was 2.7 ± 0.6° (99% confidence interval = 4.6°, maximum trajectory angle difference = 5°).

Clinical Spinal Applications

As this form of 2D neuronavigation is based on fluoroscopy, the use of a good fluoroscopic technique is necessary. Specifically, the anatomic region of interest should be placed in the center of the fluoroscopic image to minimize parallax. Furthermore, a true image should always be obtained if possible (e.g. in a true anterior-posterior (AP) view, the spinous process should bisect the pedicle). Prior to clinical use, it must be clearly understood that virtual fluoroscopy systems typically cannot improve the image quality generated by a given C-arm and it cannot compensate for a surgeon's misinterpretation of fluoroscopic (2D) osseous anatomic data. Therefore, a sound knowledge of fluoroscopic spinal anatomy is essential for the successful use of a virtual fluoroscopy system.

Virtual fluoroscopy is presently being used for the following spinal procedures: open pedicle screw placement (C7-S1), percutaneous pedicle screw placement (T10-S1), lumbar interbody cage insertion, C1-C2 transarticular screw placement, lateral mass screw placement, odontoid screw fixation and transoral decompression.

Posterior Spinal Procedures

For posterior spinal procedures, patient setup and surgical exposure are identical to the surgeon's normal technique. The DRA is rigidly affixed to the spinous process of the vertebra. For a typical degenerative case (without gross instability), the end vertebrae can be utilized for this purpose (e.g., attach the DRA to L4 for screw placement at L4, L5, and S1). The fluoroscopic views preferred by the surgeon are then acquired. These may include lateral views, AP views, or oblique (owl's eye) views down the length of the pedicle. Once processed (see System Overview), the desired images are simultaneously displayed and anatomical correlation of the tracked instrument compared to its virtual position is carried out. Utilizing virtual multiplanar fluoroscopic information, the starting point and trajectory for a given task are chosen. Clinical judgment and tactile feedback are used to refine the trajectory as necessary. These systems can provide the virtual projection of any desired length and diameter onto the chosen trajectory. The use of tracked awls, probes, and taps permits continuous visualization of the instruments along their course (fig. 2).

Percutaneous Pedicle Screws

Virtual fluoroscopy systems do not require direct exposure of the spine for registration, thus percutaneous navigation may be performed as follows. Through a small incision, the DRA is rigidly affixed to the desired spinous process. The appropriate fluoroscopic projections (AP, lateral, and oblique) are obtained and calibrated. From the skin surface, the trajectory of the tracked probe is virtually extended through the pedicle to visualize the anticipated course of the pedicle screw. Preoperative CT and MRI are used to decide the optimal entry point and trajectory for the pedicle screw. Using a cannulated

Fig. 2. AP and lateral virtual fluoroscopic views of the lumbar spine (same patient as in fig. 1). The ability to follow the course of the pedicle probe simultaneously in the AP and lateral planes enables reliable placement of the probe along an intended trajectory. Advancement of image-guided pedicle probe trough the S1 pedicle to the sacral promontory (white arrow) is demonstrated.

Fig. 2. AP and lateral virtual fluoroscopic views of the lumbar spine (same patient as in fig. 1). The ability to follow the course of the pedicle probe simultaneously in the AP and lateral planes enables reliable placement of the probe along an intended trajectory. Advancement of image-guided pedicle probe trough the S1 pedicle to the sacral promontory (white arrow) is demonstrated.

screw system the pedicle is probed and tapped under virtual fluoroscopic assistance (fig. 3) [10].

Anterior-Cervical Spine

The surface anatomy of the anterior cervical spine is typically prohibitive to anatomical registration using conventional 3D neuronavigation. Furthermore, the anatomy of the anterior cervical spine and the nature of anterior cervical spinal procedures typically do not allow the practical or safe attachment of a DRA directly to the spine. At the craniocervical junction, the anatomical relationship of the occiput-C1-C2 can often be maintained by the use of a Mayfield apparatus. As a result of the rigid relationship between the Mayfield apparatus and the occiput, a DRA directly attached to the Mayfield combined with automatic registration enables virtual fluoroscopic navigation during procedures at the craniocervical junction (e.g. odontoid screw insertion or transoral decompression). For anterior procedures of the subaxial cervical spine, the DRA can be directly attached to one or more vertebral bodies using threaded distraction pins. As noted for posterior procedures, the multiplanar position and trajectory of tracked surgical tools are provided in real time. Specific to anterior cervical procedures, information with respect to the anatomical midline and depth is also utilized.

Clinical Outcomes

The authors have significant clinical experience utilizing the FluoroNav™ system (Medtronic Surgical Navigation Technologies, Louisville, Colo., USA). In a current prospective study the clinical accuracy of virtual fluoroscopy for the placement of thoracic and lumbosacral pedicle screws is being evaluated (unpubl. data). In this study, postoperative CT of 360 pedicle screws [45 patients; lumbosacral (L1-S1) = 279; thoracic T (T3-T12) = 81] were independently reviewed by a spinal fellow and neuroradiologist. All screws were placed at an academic teaching center using the FluoroNav system. Residents and fellows placed over 50% of the screws. The relative position of the screw to the pedicle was assessed and graded as follows: A = completely in; B = <2 mm; C = 2-4mm; D = >4-6 mm. Any borderline position was automatically downgraded. If an osseous breach occurred, the direction of the breach was further classified. To date, 49 out of 360 screws (13.6%) breached the pedicle wall (fig. 4). Overall pedicle breaches were grade B in 11.9%, grade C in 1.4% and grade D in 0.3% of screws. The majority (87.5%) of breaches were minor (grade B). Overall medial and lateral breaches occurred equally. Due to anatomic constraints, breaches were 3 times more likely to occur in the thoracic spine. Thirty-five percent of breaches were secondary to a pedicle screw that was larger than the size of the pedicle. There were no clinically significant screw misplacements and no screws required revision. In this series, the potential for any neurological injury (medial pedicle breach, greater than 2 mm) was 0.6%. This study represents a worse case scenario assessment (CT analysis). The overall misplacement rate in this ongoing study is less than or comparable to reported misplacement rates using other techniques [11-13].

Fig. 3. a Percutaneous pedicle screw insertion can be accomplished utilizing multiplanar virtual fluoroscopy and an image-guided pedicle probe such as a bone biopsy needle. b As depicted, the pedicle can be pictured as a cylinder. c A percutaneously placed pedicle probe is positioned at the pedicle entry point (the 'top' of the pedicle cylinder on the lateral view and the lateral edge of the pedicle cylinder on the AP view). Its trajectory is then virtually extended to the 'bottom' of the pedicle cylinder (the junction of the pedicle and the vertebral body). By maintaining a probe trajectory that is lateral to the medial edge of the pedicle (arrow - AP view) at the 'bottom' of the pedicle, the pedicle can be percutaneously probed with minimal risk of a medial pedicle breach. When using this technique, it is imperative that true AP and lateral views are utilized. d After the pedicle is probed, a guide wire is introduced and a cannulated pedicle screw can be placed. Postoperative AP (e) and lateral x-ray (f) following a minimally invasive posterior lumbar interbody fusion augmented with the Sextant™ percutaneous pedicle screw system (Medtronic - Sofamor Danek, Memphis, Tenn., USA). g Postoperative CT image (see e, f) demonstrating ideal placement of percutaneously placed L5 pedicle screws utilizing virtual fluoroscopic assistance.

Mri After Pedicle Screw Fixation

Fig. 4. Postoperative CT images demonstrating pedicle screw placement using virtual fluoroscopic assistance. a At T6 both screws (5.5 mm) are in excellent position within the pedicles. b At T4, the left screw is in good position and the right screw demonstrates a minor breach (<2mm) of the medial pedicle wall (arrowhead). In this case the screw diameter (5.5 mm) was greater than that of the pedicle.

Fig. 4. Postoperative CT images demonstrating pedicle screw placement using virtual fluoroscopic assistance. a At T6 both screws (5.5 mm) are in excellent position within the pedicles. b At T4, the left screw is in good position and the right screw demonstrates a minor breach (<2mm) of the medial pedicle wall (arrowhead). In this case the screw diameter (5.5 mm) was greater than that of the pedicle.

Using a novel instrumentation technique, Foley et al. [10] placed percutaneous lumbar pedicle screws utilizing the FluoroNav system. Twelve patients were successfully treated using this technique. The versatility of the imaging system allowed registration of unexposed spine elements for the percutaneous procedure. Screw placement was without complication in all cases.

Although the authors only have clinical experience using the FluoroNav system, other commercially available virtual fluoroscopy systems seem to have similar accuracy and likely will demonstrate similar clinical results. In an in vitro study comparing a virtual fluoroscopy system to a conventional 3D image-guided technique, Choi et al. [7] demonstrated comparable accuracy. The overall screw misplacement rate (17.9%) demonstrated in this in vitro study is comparable to the clinical findings of Rampersaud et al. (see above, unpubl. data). Based on extensive in vitro data and in vivo experience the authors have concluded that FluoroNav appears to be a safe adjunct for the placement of thoracic and lumbosacral pedicle screws.


Although virtual fluoroscopy has many advantages over conventional fluoroscopy, its primary limitation is 2D image generation. Also, the quality of the fluoroscopically generated images and the surgeon's ability to meaningfully interpret these views are unchanged compared to conventional fluoroscopy.

For all current image guidance systems, the navigational accuracy is greatest at the spinal segment to which the DRA is attached. The rigid relationship between the DRA and the spinal segment to which it is attached allows for accurate tracking and compensation for any motion occurring at that specific spinal segment. As a result of intersegmental motion and other technical reasons, overall navigational accuracy decreases as one moves further away from the DRA. Consequently, to maintain maximal navigational accuracy, the DRA should be moved to each spinal segment of interest with new fluoroscopic images acquired at each segment. This, however, is clinically impractical for multilevel cases. Unless gross segmental spinal instability is present, acceptable clinical accuracy is typically maintained for 2-5 spinal segments for one DRA position (1 segment: above, 1: at and 1-3: below). However, clinical judgment must be used to determine whether the system is adequately correlating to the segment of interest before navigation is carried out. For spinal procedures such as C1-2 transarticular screws where a low margin of error exists the DRA must be attached to the vertebrae of interest. Furthermore, by obtaining a live image to confirm the position of a tracked instrument the fluoro-scope can serve as its own internal validation system.

Although virtual fluoroscopy provides the advantages of intraoperative imaging, it is still limited by the inability to easily track multiple spinal segments. Therefore, it is still necessary (for the reasons mentioned above) to move the DRA and acquire new fluoroscopic images several times during long segment procedures. The overall number of C-arm positional changes and images required, however, is still much less when compared to conventional C-arm fluoroscopy.

Under ideal circumstances surgical navigational systems can be very accurate. However, there are numerous potential sources of error that can significantly affect the clinical accuracy of any image guidance system. These include errors related to imaging, tracking, registration algorithm(s), and the tracked surgical tools. These errors, although typically submillimetric, are cumulative and can result in a clinically significant overall navigational error. One of the potentially largest, but often understated sources of errors is the surgeon. Like any other tool, a virtual fluoroscopic navigational system is only as good as its user. Sound knowledge of the limitations associated with these systems is paramount for their optimal use.

Future Implications

The future of virtual fluoroscopy is one of versatility. The practical implementation of an easy to use intraoperative imaging modality combined with surgical navigational technology has enabled a robust platform from which only improvement is possible. Virtual fluoroscopy has the potential to bring practical 3D intraoperative image guidance to the operating room. This is currently evolving in two pathways. The first is a direct method utilizing a novel isocen-tric C-arm (Siemens). This C-arm is capable of generating a 3D image data set, therefore allowing 3D neuronavigation without the hassles of 3D registration. Furthermore, because it is an intraoperative imaging device certain limitations such as positional (patient) intersegmental motion and intraoperative image update of conventional 3D neuronavigation are eliminated. The second pathway involves technology that allows automated registration of intraoperative 2D fluoroscopic images to a preacquired 3D image data set [14]. The goal of both techniques is to eliminate the time-consuming step of tactile anatomic registration, while providing unparalleled 3D anatomic information and practical intraoperative imaging capabilities. 2D to 3D image fusion also allows incorporation of multiple imaging modalities (e.g. CT, MRI, ultrasound or angiographic). The added anatomical detail of a 3D image data set without the need for tactile registration is ideally suited for the refinement and development of percutaneous or minimal access spinal surgery.

Virtual fluoroscopy is also facilitating another exciting area of research that involves practical solutions for tracking multiple segments of the spine. Combined with segmentation of an image data set this will allow the user to maintain optimal navigational accuracy over numerous segments and update the relative position of individual segments in real time. Finally, it is only logical that this technology will likely be integrated directly into the C-arm fluoroscope, thus providing a versatile intraoperative imaging-navigational system.


Virtual fluoroscopy is a safe and beneficial adjunct to complex spinal surgery. The use of this technology has shown improved or comparable pedicle screw misplacement rates when compared to published results utilizing anatomical landmarks with or without fluoroscopy and conventional 3D image guidance, respectively. As with all neuronavigation devices, advantages and disadvantages specific to virtual fluoroscopy exist and must be considered in the decision to utilize this technology. The real-time multiplanar information provided by these systems can be a powerful adjunct to the surgeon's ability to perform complex tasks during spinal surgery. However, it must never be forgotten that this technology is intended to serve as an adjunct to the surgeon's clinical judgment and technical skills, not a replacement.


1 Foley KT, Smith MM: Image-guided spine surgery. Neurosurg Clin N Am 1996;7/2:171-186.

2 Rampersaud YR, Foley KF: Image-guided spinal surgery. Operative Tech Orthopaed 2000;10/1: 64-68.

3 Rampersaud YR, Foley KT, Shen AC, Williams S, Solomito M: Radiation exposure to the spine surgeon during fluoroscopically assisted pedicle screw insertion. Spine 2000;25:2637-2645.

4 Foley KT, Rampersaud YR, Simon DA: Virtual fluoroscopy: Multi-planar x-ray guidance with minimal radiation exposure. Eur Spine J 1999;8(suppl 1):S36.

5 Nolte LP, Slomczykowski MA, Berlemann U, Strauss MJ, Hofstetter R, Schlenzka D, Laine T, Lund T: A new approach to computer-aided spine surgery: Fluoroscopy-based surgical navigation. Eur Spine J 2000;9(suppl 1):S78-S88.

6 Foley KT, Simon DA, Rampersaud YR: Virtual fluoroscopy. Operative Tech Orthopaed 2000;10/1: 77-81.

7 Choi WW, Green BA, Levi ADO: Computer-assisted fluoroscopic targeting system for pedicle screw insertion. Neurosurgery 2000;47:872-878.

8 Foley KT, Simon DA, Rampersaud YR: Virtual fluoroscopy: Computer-assisted fluoroscopic navigation. Spine 2001;26:347-351.

9 Jane JA, Thapar K, Alden TD, Laws ER Jr: Fluoroscopic frameless stereotaxy for transsphenoidal surgery. Neurosurgery 2001;48:1302-1308.

10 Foley KT, Gupta SK, Justis JR, Sherman MC: Percutaneous pedicle screw fixation of the lumbar spine. Neurosurg Focus 2001;10/4:1-8.

11 Merloz P, Tonetti J, Pittet L, Coulomb M, Lavallee S, Sautot P: Pedicle screw placement using image guided techniques. Clin Orthop 1998;354:39-48.

12 Laine T, Lund T, Ylikoski M, Lohikoski J, Schlenzka D: Accuracy of pedicle screw insertion with and without computer assistance: A randomised controlled clinical study in 100 consecutive patients. Eur Spine J 2000;9:235-240.

13 Youkilis AS, Quint DJ, McGillicuddy JE, Papadopoulos SM: Stereotactic navigation for placement of pedicle screws in the thoracic spine. Neurosurgery 2001;48:771-778.

14 Theodore N, Sonntag VKH: Spinal surgery: The past century and the next. Neurosurgery 2000;46: 767-776.

Y. Raja Rampersaud MD, FRCSC

Toronto Western Hospital, University Health Network

Edith Cavell Wing, 1-039, 399 Bathurst St., Toronto, ON M5T 2S8 (Canada)

Tel. +1 416 603 5399, Fax +1 416 603 3437, E-Mail [email protected]

Haid RW Jr, Subach BR, Rodts GE Jr (eds): Advances in Spinal Stabilization. Prog Neurol Surg. Basel, Karger, 2003, vol 16, pp 96-106

Back Pain Relief

Back Pain Relief

This informational eBook will present you with the most recent research and findings available so that you can learn more about Back Pain relief, covering as many bases as possible from A to Z.

Get My Free Ebook

Post a comment