Virtual Fluoroscopy: Multiplanar X-Ray Guidance with Minimal Radiation Exposure

David A. Simon, Ph.D.
Medtronic Surgical Navigation Technologies, Broomfield, Colorado

Y. Raja Rampersaud, M.D.
Division of Neurosurgery and Orthopaedic Surgery, University of Toronto, Toronto, Ontario, Canada

Abstract

Study Design: In vitro accuracy assessment of a novel virtual fluoroscopy system.

Objectives: To investigate a new technology combining image–guided surgery with C–arm fluoroscopy.

Summary of Background Data: Fluoroscopy is a useful and familiar technology to all musculoskeletal surgeons. Its limitations include radiation exposure to the patient and operating team and the need to reposition the fluoroscope repeatedly to obtain surgical guidance in multiple planes.

Methods: Fluoroscopic images of the lumbar spine of an intact, unembalmed cadaver were obtained, calibrated, and saved to the StealthStation. The FluoroNaV(tm) system was used for the sequential insertion of an LED–fitted probe into the pedicles of Ll to S1 bilaterally. The trajectory of a "virtual tool" corresponding to the tracked tool was overlaid onto the saved fluoroscopic views in real–time. Live fluoroscopic images of the inserted pedicle probe were then obtained. Distances between the tips of the virtual and fluoroscopically–displayed probes were quantified using the image–guided computer's measurement tool. Trajectory angle differences were measured using a standard goniometer and printed copies of the StealthStation display. Surgeon radiation exposure was measured using thermolucent dosimeter (TLD) rings.

Results: Excellent correlation between the virtual fluoroscopic images and live fluoroscopy was observed. Mean probe tip error was 0.97 mm +/– 0.40 mm. Mean trajectory angle difference between the virtual and fluoroscopically–displayed probes was 2.7 +/– 0.6 degrees. The TLD rings measured no detectable surgeon radiation exposure.

Conclusions: Virtual fluoroscopy offers several advantages over conventional fluoroscopy while providing acceptable targeting accuracy. It enables a single C–arm to provide real–time, multiplanar procedural guidance. It also dramatically reduces radiation exposure to the patient and surgical team by eliminating the need for repetitive fluoroscopic imaging for tool placement.

Introduction

Fluoroscopy is routinely used for intraoperative localization of patient anatomy and surgical instrument position. By providing this information, it facilitates improved accuracy and reduced surgical exposure for a wide variety of procedures. Furthermore, fluoroscopy has enabled the development of many interventional and surgical techniques, such as intramedullary nailing of long bone fractures.

The use of fluoroscopy is familiar to most surgeons, particularly in trauma management. It can also be helpful in a variety of spinal procedures. Odontoid screw and interbody cage placement are facilitated by fluoroscopy. The development of a variety of percutaneous spinal procedures, such as vertebroplasty, has depended on the use of fluoroscopy. In addition, many spine surgeons routinely use fluoroscopic assistance for the placement of pedicle screws.

Despite its widespread acceptance and utility, however, fluoroscopy does have disadvantages. The most notable is occupational radiation exposure, particularly to the surgeon's hands. (4,5,8) Recent data suggest that spinal surpeons, in particular, are at significant risk for fluoroscopy–related radiation exposure.(6)

There are also other limitations associated with the use of a mobile C–arm fluoroscope. For example, only a single real–time planar view is usable at any given time. Consequently, for procedures requiring multiplanar fluoroscopic visualization, such as percutaneous transpedicular biopsy, the C–arm has to be repositioned throughout the procedure. This process is often tedious, time–consuming, and frustrating. Under ideal circumstances the surgeon holds an instrument perfectly still in one plane while correcting its position in the other. The x–ray technologist efficiently repositions the C–arm to obtain the perfect view in each desired plane while maintaining ideal sterility and without ergonomically challenging the surgeon with the C–arm. Unfortunately, this scenario is definitely the exception, not the rule.

By combining current C–arm fluoroscopy with computer–aided surgical technology, many advantages of fluoroscopy can be enhanced, while minimizing or eliminating its disadvantages. The purpose of this study was to investigate a new technology combining image–guided surgery and C–arm fluoroscopy, which we term "virtual fluoroscopy."

Methods

There are four fundamental steps required for a virtual fluoroscopy system: 1) transferring one or more fluoroscopic images to a computer for subsequent processing; 2) measurement of the spatial relationship between the images and the patient; 3) "calibration" of the images acquired in Step 1; and 4) superimposition of a "virtual surgical instrument" onto the acquired images using measurements of the actual instrument's position, together with the information gathered in Steps 1–3.(1) In Step 1, a fluoroscopic image is transferred from the fluoroscope to the computer via a standard video connection or a digital link. In Step 2, the relative position of the C–arm and the patient at the time of image acquisition is measured. This is typically done using a position–measuring sensor that uses special cameras (e.g., an electro–optical camera) to accurately track the positions of LEDs (light–emitting diodes) or passive reflectors that have been affixed to the C–arm and the patient. These markers are affixed to the patient using a dynamic reference array (DRA), which rigidly attaches to the anatomy to be navigated. In Step 3, the computer calibrates the acquired image using information from the position measurement from Step 2. During the calibration process, the computer builds a mathematical description of the fluoroscopic image formation process. This description geometrically relates how a given spatial position relative to the patient projects into the fluoroscopic image and conversely, how a given picture element of the fluoroscopic image projects back through the patient to the fluoroscope's radiation source. This geometric description may be different for every acquired image due to such factors as variations in the C–arm's position relative to the Earth's magnetic field, the effect of external electromagnetic fields in the vicinity of the C–arm generated by other electrical devices in the operating room, and the mechanical non–rigidity of the C–arm structure. For this reason, it is critical that every acquired image be independently calibrated in a virtual fluoroscopy system. Finally, in Step 4, the computer measures the position of one or more surgical instruments using the position–measuring sensor, and generates a graphical overlay that indicates precisely where the surgical instrument would appear in the fluoroscopic image if the image were being continuously updated. In this manner, the computer can continuously superimpose the position of the surgical instrument on one or more previously–acquired fluoroscopic images without additional radiation exposure to the patient or surgical team.

In this study, the FluoroNaV(tm) Virtual Fluoroscopy System (Medtronic Surgical Navigation Technologies, Broomfield, CO) was used. A calibration target with affixed light–emitting diodes (LEDs) was attached to an OEC Model 9600 C–arm fluoroscope (OEC, Salt Lake City, UT). The StealthStation image–guided surgery system was used to track the fluoroscope as well as a dynamic reference array and various spine surgery tools. Fluoroscopic images of the lumbar spine of an intact, unembalmed cadaver were obtained, calibrated, and saved to the StealthStation. The surgeon stood at least six feet away from the fluoroscope's x–ray source during image acquisition. The trajectory of a "virtual tool" corresponding to the tracked tool was overlaid onto the saved fluoroscopic views in real–time. Sequential insertion of an LED–fitted probe into the cadaver's pedicles from LI to SI bilaterally was performed under virtual fluoroscopic guidance. Live fluoroscopic images of the inserted pedicle probe were then obtained; distances between the probe tips and differences between the trajectory angles of the virtual and fluoroscopically–displayed probes were measured. Angular measurements were made using a standard goniometer and printed copies of the StealthStation display. All linear measurements were made directly utilizing the image–guided computer's measurement function. Surgeon radiation exposure was measured using thermolucent dosimeter (TLD) rings. The study was performed in a mock operating room setting using a standard electrical operating room table, standard positioning and draping, and standard surgical techniques.

Results

All pedicles were successfully probed with virtual fluoroscopic guidance. Excellent correlation between the virtual fluoroscopic images and live fluoroscopy was observed. Mean probe tip error was 0.97 mm +/– 0.40 mm. The 99% confidence interval was 2.2 mm. Mean trajectory angle difference between the virtual and fluoroscopically–displayed probes was 2.7 +/– 0.6 degrees. The 99% confidence interval was 4.6 degrees. Maximum probe tip error was 3 mm; maximum trajectory angle difference was 5 degrees. The TLD rings measured no detectable surgeon radiation exposure.

Discussion

A virtual fluoroscopy system for spinal and musculoskeletal procedures offers several distinct advantages over conventional C–arm fluoroscopy. First, radiation exposure to the patient and surgical team is reduced. The system eliminates the need to take multiple images to update instrument position; rather, the instrument is tracked by the digitizer and its real–time position is overlaid onto the previously acquired fluoroscopic view(s). In addition, bilateral localization at any given spinal level(s) can be performed using a single image, further reducing fluoroscopy time. Furthermore, as pre–acquired fluoroscopic images are used for navigation, the surgical team can stand at a safe distance during "live" fluoroscopy, minimizing or eliminating the need for wearing lead shielding. Second, a single C–arm unit is turned into a multiplanar device. The surgeon can pre–acquire several images in several planes and use them for navigation. The system overlays a tracked tool's position onto all of the pre–acquired views simultaneously (up to four views). Thus, virtual fluoroscopy eliminates the need to repeatedly reposition the C–arm and enables the surgeon to achieve a desired trajectory in a much more efficient manner. Third, after acquiring the desired images, the surgeon can move the C–arm out of the operative field, minimizing or eliminating the ergonomic challenges of C–arm use, particularly in spinal procedures. Fourth, the computational power of the image–guided computer allows further enhancement of standard fluoroscopy by providing real–time quantitative information to the surgeon. For example, in planning pedicle screw insertion, the distance of the screw insertion point from the midline and the desired axial trajectory can be obtained from the patient's preoperative CT or MRI (e.g., the diagnostic study, not a specially–formatted image– guided study). After obtaining a true AP fluoroscopic view and defining the midline with the FluoroNaV(tm) software, the surgeon can see a real–time numeric display of the angular trajectory of an instrument relative to the mid–sagittal plane (in degrees) and the distance of its tip from the midline (in millimeters).

This study demonstrates acceptable in vitro accuracy using a virtual fluoroscopy system that is similar to the overall accuracy reported for three–dimensional (3–D) image–guided surgery systems. (2,3,7) Compared to current 3–D image–guided surgery systems, virtual fluoroscopy is readily applicable to a wide variety of musculoskeletal and spinal procedures and has several advantages. The time and cost of obtaining specially formatted preoperative CT or MR imaging is avoided. The often time–consuming and frustrating step of image to patient registration is unnecessary; in fact, the image calibration process for virtual fluoroscopy is fully automated. In addition, real–time image updating for positional changes of the patient or after manipulation of a given spinal segment is easily achieved intraoperatively by simply acquiring a new fluoroscopic image. Likewise, real–time intraoperative fluoroscopic validation of the virtually displayed instrument position can be obtained at any time, providing a "safety check."

Despite these many advantages, it must be stated that virtual fluoroscopy is a two–dimensional (2–D) navigational system. It does not provide the detailed multiplanar imaging generated by 3–D systems and, consequently, has different clinical utility. Errors in the clinical interpretation of 2–D images and the extrapolation of 2–D information to 3–D anatomy are still dependent on the expertise of the spine surgeon. Furthermore, inherent to the nature of fluoroscopy, other disadvantages remain. A virtual fluoroscopy system cannot compensate for causes of poor image quality such as obesity, bowel contrast following trauma assessment, or the inadvertent positioning of radio–opaque structures (e.g., refractors or patient positioning devices). This system is adaptable to other digital fluoroscopes; however, image quality can vary significantly among different fluoroscope models. The effects of parallax also need to be considered when using a virtual fluoroscopy system. The use of proper fluoroscopic techniques such as centering the area of interest within the field of view and tangential imaging (acquiring "true" lateral and AP views) still need to be adhered to. Clinical misinterpretation of a low quality, poorly oriented image cannot be compensated for by any navigational system.

This is a limited study that presents preliminary data on the utility and in vitro accuracy (i.e., correlation to conventional fluoroscopy) of a virtual fluoroscopy system. Current in vitro and clinical studies assessing the accuracy of this system in multiple planes and in different clinical applications are promising. The authors (KTF, YRR) have been encouraged with its clinical utility and accuracy and are part of an ongoing, multicenter clinical trial.

In conclusion, virtual fluoroscopy expands upon the real–time imaging advantages of conventional fluoroscopy, while reducing or eliminating many of its disadvantages. Computer enhancement of fluoroscopically–assisted procedures provides a broad–based and practical application of surgical navigational technology.

References

1. Foley KT, Simon DA, Rampersaud YR. Virtual Fluoroscopy. Operative Techniques in Orthopaedics 2000; 10:77–81.

2. Glossop ND, Hu RW. Practical accuracy assessment of image guided spine surgery. Presented at the 2nd Annual North American Program on Computer– Assisted Orthopaedic Surgery, Pittsburgh, Pennsylvania, June 1998.

3. Helm PA, Eckel TS. Accuracy of registration methods in frameless stereotaxis. Comput Aided Surg 1998; 3:51–6.

4. International Commission on Radiological Protection, Publication 60: Recommendations of the International Commission on Radiological Protection. Ann ICRP 1991; 21:1–3.

5. Mehlman CT, DiPasquale TG. Radiation exposure to the orthopaedic surgical team during fluoroscopy: "How far away is far enough?" J Orthop Trauma 1997; 11:392–398.

6. Rampersaud YR, Foley KT, Shen AC, Williams S, Solomito M. Radiation exposure to the spine surgeon during fluoroscopically–assisted pedicle screw insertion. Spine (accepted for publication).

7. Simon DA, O'Toole RV, Blackwell M, et al. Accuracy validation in image–guided orthopaedic surgery. Proceedings of the 2nd International Symposium on Medical Robotics and Computer–Assisted Surgery 1995; 185–192.

8. Sanders R, Koval KJ, DiPasquale T, Schmelling G, Stenzier S, Ross E. Exposure of the orthopaedic surgeon to radiation. J Bone Joint Surg 1993; 75–A:326–330.

*Text from: Foley KT, Simon DA, Rampersaud YR. Virtual fluoroscopy: multiplanar x–ray guidance with minimal radiation exposure. (Submitted for publication)

The authors would like to thank the Medical Education & Research Institute for its support of this project. Financial support for this project was provided by Medtronic Sofamor Danek.