Fracture-table-mounted versus bone-mounted dynamic reference frame tracking accuracy using computer-assisted orthopaedic surgery–a comparative study

Abstract

Objective: Fluoroscopy-based computerized navigation systems enable accurate implant placement while reducing radiation exposure. The navigation process normally requires the attachment of a dynamic reference frame (DRF) to a bone, causing additional surgical trauma. The aim of this study was to compare the accuracy of navigation with the DRF either attached to the bone or mounted on the fracture table.

Methods: We conducted a prospective study on 10 consecutive patients who underwent operative fixation of femoral neck fractures with cannulated screws using computerized navigation. After insertion of the three guide wires, the DRF was moved from the patient's bone to the fracture table. For each screw, angular and translational deviations of the navigated images as compared to the conventional fluoroscopic images were analyzed.

Results: The accuracy of navigated Kirschner wire placement was similar with both techniques, resulting in an average translational error of less than 2 mm in both groups and around 1° in angulation error–both of these accuracy measurements are acceptable and sufficient for the insertion of cannulated screws into the femoral head.

Conclusion: Our study suggests that attaching the DRF to a fracture table during navigated femoral neck fixation allows for acceptable accuracy with the possible added benefit of reducing patient morbidity.

• Computer assisted orthopaedic surgery
• navigation
• dynamic reference frame

Introduction

Fluoroscopy-based computerized navigation systems have been introduced for clinical use in orthopaedic traumatology applications [1–4]. These systems enable the surgeon to track virtual images of surgical instruments superimposed on previously acquired multiplanar fluoroscopic images, an approach known as virtual or augmented fluoroscopy. Fluoroscopy-based computerized navigation uses a tracking unit comprised of a computer, a position sensor (typically an infrared optical camera), and a calibration target placed on a C-arm fluoroscope. To account for motions, a rigidly fixed tracker called a dynamic reference frame (DRF) must be affixed to a bony structure near the surgical site. Usually, this is done by drilling one or more screws or pins into a bone adjacent to the surgical site via a separate surgical incision. The DRF is then connected to the screws or pins (Figure 1). Following acquisition of fluoroscopic images with the DRF in place, navigation can begin. During navigation, the surgical instrument is tracked by triangulation with the DRF and the position sensor. Virtual images of the instrument are then projected on the previously acquired images without the need for additional fluoroscopy (Figure 1). In fact, the process of navigation is equivalent to continuous intraoperative multiplanar fluoroscopic imaging without the use of the actual fluoroscope. Several studies have validated the accuracy and reliability of the navigation systems, as well as the significant reduction in radiation exposure resulting from their use [1], [2], [5], [6].

Figure 1. Standard fluoroscopic-based navigation for cannulated screw insertion. The dynamic reference frame (labeled RF) is attached to the iliac crest. A tracked drill guide (DG) is used for planning the screw's trajectory, and a guide wire is than inserted through it. The computer display (CD) displays the virtual image of the drill guide trajectory (green line) on both the AP and lateral views simultaneously. [Color version available online.]

Cannulated screw fixation is a common orthopaedic procedure for the surgical fixation of femoral neck fractures. Accuracy in screw placement contributes to the success of this procedure [7–12], and the use of computerized navigation can facilitate accurate placement of the implants within the femoral neck [2], [4], [13].

Although initial results with navigated screw insertion are encouraging [13], the insertion of the DRF involves several problematic aspects: (1) its insertion into a non-involved bone creates an additional (albeit small) operative site in which, subsequently, local wound complications and bony damage may occur, especially with osteoporotic patients; (2) the DRF can cause interference with other surgical instruments; and (3) due to stability considerations, the DRF may have to be mounted far away (> 20 cm) from the surgical site, thereby decreasing the accuracy of navigation.

As an alternative, we propose to modify the navigated procedure by attaching the DRF to the fracture table instead of to the bone. Our hypothesis is that for certain trauma procedures, the accuracy of the navigated procedure can be maintained using the fracture table as a base for mounting the DRF. Possible additional applications include insertion of percutaneous fixation screws for SI joint disruptions and the starting point for proximal femoral nailing.

The purpose of this study was to compare the radiographic accuracy of navigation when the DRF was placed on the fracture table versus the patient's bone.

Materials and methods

Patients

The study population consisted of 10 consecutive patients with femoral neck fractures (AO/OTA types 31B1, 31B2.1) who were operated during the period from November 2004 to March 2005 using the standard navigated technique described elsewhere [4], [13]. Intraoperative measurements were used for the purposes of the study.

Surgical technique

The iON Fluoronav® StealthStation® navigation system (Medtronic SNT, Louisville, CO) was used as the image guidance platform for the surgery. The patients were secured to a fracture table in a standard fashion and closed reduction was performed when necessary. Both lower extremities were bound to the fracture table leg holders. An additional plastic post (PORD™, Efratgo, Haifa, Israel) was placed under the proximal femur to provide additional support. After preparation and draping, a DRF was attached to the involved side iliac crest using two 3-mm Schanz screws. Anteroposterior (AP) and lateral radiographs of the involved hip were registered with the tracked C-arm fluoroscope. Three 3.2-mm guide wires (threaded Kirschner wires) were inserted into the femoral neck using the standard fluoroscopy-based navigation technique, as described in reference [13] (Figure 1).

Until now, this has been the routine surgical procedure used for fixing femoral neck fracture in our institution [13]. To determine whether the information given by the navigation system varies following relocation of the DRF to the table, the following measures were taken:

1. New radiographs of the hip following implantation of the three guide wires were acquired. The cannulated tracked drill guide was re-positioned over the guide wires. Therefore, the navigated (“virtual”) image was superimposed on the actual (“true”) image of the implanted guide wire (Figure 2). This technique was described previously by Kahler et al. [4] for validating accuracy. Ideally, when no relative motion occurs, these two images should overlap. In practice, however, a small error usually occurs due to deformation of the guide wire and some motion while performing the procedure. This error served as a basis for comparison between the two techniques studied (see below).

2. The images obtained in the above validation of each of the three guide wires on both AP and lateral images were captured and saved as bitmap images in the computer memory.

3. The DRF was subsequently moved to the new location; it was connected to the table by means of a modified external fixator frame (Figure 3). New AP and lateral images were then acquired, and the process described in steps 1 and 2 above was then repeated with the table-fixed DRF.

Figure 2. The validation phase. New images are taken after the insertion of the guide wires. The drill guide is placed over the wires and the navigated virtual image (green line) is superimposed on the “real” radiographic image (black line). The discrepancy between these images in both translation and angulation is calculated. [Color version available online.]

Figure 3. Table-mounted (A) and iliac-crest–mounted (B) dynamic reference frame attachments. The drill guide (C) is also visible.

For every guide wire a total of four images were saved in the computer memory: one AP and one lateral projection image with a virtual trajectory superimposed using the bone-mounted DRF, and the same images using the table-mounted DRF (a total of 12 images for each case). At the end of this process, the screws were inserted over the guide wires using the standard technique. It should be stressed that the actual guide wires were inserted using the standard technique and the alternative reference frame was used for radiographic measurements only. All cases and intraoperative navigation were performed by a single surgeon (Y.A.W.).

The images were downloaded and analyzed by a single evaluator (the first author, I.I.) using Adobe Photoshop Version 5.0 Limited Edition software (Adobe Systems, Inc., San Jose, CA). Each image underwent size calibration using the known computerized trajectory length as the scaling reference. On each image, the coordinates of the center of the navigated trajectory and of the drill guide were obtained at two different points, and the equation for the straight line representing the center was derived. From the centerline of the drill guide, the trajectory's translation was measured at two sites: the entry point (d_entry) and the trajectory's tip (d_tip). The angle between the trajectory image and the k-wire fluoroscopic images (alpha) was also calculated (Figure 4).

Figure 4. Determination of the trajectory translation at the entry point (d_entry) and tip (d_tip), and of the angle between the trajectory image and the k-wire fluoroscopic images (alpha).

A total of 29 screws were available for analysis (in one case the screw image was not available due to technical problems). A two-tailed t-test was used for statistical analysis, with a p-value of 0.05 or less being considered statistically significant.

Results

For the AP view, with the reference frame attached to the iliac crest, the average translational deviation of the trajectory from the inserted guide wire was 1.18 ± 0.92 mm at the entry site and 1.25 ± 1.53 mm at the trajectory tip. When the DRF was attached to the fracture table, the average deviations were 1.24 ± 0.90 mm and 1.85 ± 1.37 mm, respectively. The differences were not statistically significant (Table I).

Table I.  Translational offset (in mm) of the trajectory from the drill guide at the two sites (d_tip and d_entry) in the AP view.

	Anteroposterior view
	Difference		Fracture table		Iliac crest
	d_tip	d_entry	d_tip	d_entry	d_tip	d_entry
Average	1.75	0.91	1.85	1.24	1.25	1.18
Min	0.03	0.00	0.00	0.00	0.00	0.00
Max	4.81	2.55	5.17	2.93	6.87	3.91
SD	1.41	0.70	1.37	0.90	1.53	0.92
p value	NS	NS

For the lateral view, when the DRF was attached to the iliac crest, the average translational deviation of the trajectory from the inserted guide wire was 1.42 ± 0.88 mm at the entry site and 1.63 ± 1.25 mm at the trajectory tip. When the DRF was attached to the fracture table, the average deviations were 1.26 ± 0.71 mm and 1.57 ± 0.85 mm, respectively (Table II). These differences were also not statistically significant.

Table II.  Translational offset (in mm) of the trajectory from the drill guide at the two sites (d_tip and d_entry) in the lateral view.

	Lateral view
	Difference		Fracture table		Iliac crest
	d_tip	d_entry	d_tip	d_entry	d_tip	d_entry
Average	1.20	1.00	1.57	1.26	1.63	1.42
Min	0.18	0.05	0.07	0.00	0.00	0.00
Max	2.86	2.11	3.03	2.62	4.87	3.02
SD	0.73	0.62	0.85	0.71	1.25	0.88
p value	NS	NS

For the AP view, the angular differences were 0.88 ± 0.82° in the group with the DRF mounted on the iliac crest and 1.07 ± 0.82° in the group with the table-mounted DRF, which is also not statistically significant. For the lateral view, angular differences were 1.05 ± 0.84° in the group with the DRF mounted on iliac crest and 1.20 ± 0.80° in the table-mounted DRF group, which are not statistically significant (Table III).

Table III.  Angular offset (in degrees) in the AP and lateral views.

	AP view		Lateral view
	Fracture table	Iliac crest	Fracture table	Iliac crest
Average	1.07	0.88	1.20	1.05
Min	0.00	0.00	0.00	0.00
Max	2.99	3.45	3.07	2.92
SD	0.82	0.82	0.80	0.84
p value	NS	NS	NS	NS

Discussion

Computerized navigation guidance systems have proven their accuracy in several clinical applications while reducing the overall amount of radiation exposure [3–6]. However, navigation systems rely on fixation of a dynamic reference frame (DRF), also termed a rigid body or bone tracker, to the patient's anatomy. This may pose three main difficulties: First, an additional surgical site is created and additional holes are drilled into the patient's bones, increasing the surgical trauma. Second, if other instruments such as navigated jigs and other tools or robots are to be mounted, the dependence on a bony tracker may limit their use. Third, due to stability considerations, the tracker may have to be mounted far from the surgical site, thereby decreasing the accuracy of the navigation.

Our first postulation was that while the patient is secured to the fracture table, the DRF can be moved from the femur, as described in earlier techniques [4], to the iliac crest, assuming no motion occurs within the hip joint. This had already proven to be the case [13], and we use this technique routinely. The next step was to assume negligible motion of the patient on the fracture table, enabling us to attach the DRF to the table. Our results demonstrate that the average translational error using both techniques is 1.25–1.85 mm at the trajectory tip, which is an acceptable range of error for the purpose of cannulated screw insertion into the femoral head. Thus, the shift of the DRF to the table did not compromise accuracy.

There is a potential advantage in the table-mounted technique of placing the reference frame close to the operated anatomy. By doing so, the average translational error decreases, since the distances between the surgical instruments, operated bone, and DRF are smaller [14]. However, a standard external fixator with a relatively long arm was used to fix the DRF to the table (Figure 3), and this long arm had some micromotion due to leverage. This is a possible explanation as to why better results using the table-mounted frame were not achieved. Future custom-made attachments should limit this motion and will increase the accuracy of table-mounted trackers. Another drawback of connecting the DRF to the table is the absolute necessity for no patient motion on the operating table. This would be problematic to achieve in cases in which motion of the extremity is necessary, such as ACL reconstruction or total joint reconstructions.

A further possible benefit of the table-mounted technique is the future application of other tracked instruments fixed to the tables, such as robots, drills, saws, etc., without the need to fix them to the patient and thereby cause major technical difficulties and possible surgical morbidity.

We conclude that for performing computer-assisted surgical tasks such as hip fracture pinning on a fracture table, the table-mounted DRF is at least as accurate as the conventional bone-mounted reference tracking frame, while reducing the surgical morbidity.