The influence of metal artifacts on navigation and the reduction of artifacts by the use of polyether-ether-ketone

Abstract

Registration is a crucial step in navigation assisted surgery. When performing anatomical pair-point registration, there are several potential sources of error, including inadequate data acquisition, improper segmentation, and distortion resulting from metal artifacts. The aim of this study was to evaluate the influence of metal artifacts on the precision of Iso-C^3D and fluoroscopy-based navigation, and to assess any changes in precision from the use of a newly developed Schanz screw composed of polyether-ether-ketone (PEEK OPTIMA®).

A T-shaped test specimen was manufactured from synthetic bone material. It was then scanned with a Siremobil® Iso-C^3D while different types of implant were present in the specimen. Five Iso-C^3D scans were acquired: one with a steel Schanz screw in the specimen, one with a titanium screw, one with a PEEK screw, one with a 5-hole plate, and one with no screw or plate present. The registration was analyzed by “reverse verification” with a pointer in a purpose-built, manipulable 3D holder. All experiments were then repeated using fluoroscopy-based navigation.

Increasing presence of metal in the scan area resulted in an increase in mean error (0.55 mm with the steel Schanz screw, 0.7 mm with the 5-hole plate). Artifacts resulting from the titanium Schanz screw were less than those caused by the stainless steel Schanz screw. While this study demonstrates that metallic artifacts do have an influence on the precision of Iso-C^3D navigation, such artifacts were not found to be a factor when performing fluoroscopy-based navigation.

• CAOS
• Iso-C^3D navigation
• sources of error
• artifacts
• PEEK
• precision

Introduction

Iso-C^3D navigation has become an important tool for computer assisted surgery, offering several advantages. The isocentric design of the system ensures that the central ray is located in the center of rotation of the C-bow at run time in all projections [1]. Furthermore, an integrated engine allows automatic orbital movements [1] in a manner that enables the equipment to run scans automatically up to 190° around an object located in the isocenter.

This results in a 3D volume data set with an edge-length of 12 cm. This intraoperatively obtained 3D data set from the Siremobil® Iso-C^3D can subsequently be translated to Iso-C^3D-based navigation via the Navi-Link® interface (Siemens Healthcare, Erlangen, German) [2], [3]. During data acquisition with the Iso-C^3D, markers located on the C-bow allow the localizer on the navigation system to determine its position in space and any subsequent movements. This is what allows the navigation system to generate the position and orientation of the 3D data set in space. The value of this device has been demonstrated in several procedures, including the drilling of osteochondral lesions in the talus [4], the placement of pedicle screws for spine surgery [5], sacroiliac screw placement in the pelvis [6], and tumor surgery [7]. As with most navigation systems, registration and data acquisition are important steps in Iso-C^3D-based navigation, and are subject to several potential sources of error. Studies have shown that the precision of navigation degrades significantly as distance of the reference base from the isocenter increases [8]. However, placement of the reference base too close to the optical path causes extensive artifacts, which can complicate the localization of the area of interest.

In a study by Kotsianos et al. [9], the comparative influence of metal artifacts on the image quality of Iso-C^3D and CT-based data was evaluated. Sixteen human cadaveric knees were reduced with various stainless steel and titanium plates and screws. The authors found that titanium caused the least artifacts while stainless steel caused the most. While the image quality of data obtained with Iso-C^3D was inferior to that of CT-acquired images, the metal artifacts were clearly more prominent in CT images as compared to those acquired with Iso-C^3D. Despite the presence of artifacts, the image quality was sufficient for diagnostic and clinical purposes.

The aim of this study was to assess the influence of metal artifacts on the precision of fluoroscopy- and Iso-C^3D-based navigation, and to determine any change in precision resulting from the use of a newly developed polyether-ether-ketone (PEEK OPTIMA®) Schanz screw.

Materials and methods

A T-shaped test specimen was manufactured from synthetic bone material (Synbone, Switzerland). It consisted of two separate pieces joined together: The first piece, which served as the horizontal portion of the T, measured 92 mm × 50 mm and was used to represent the region of interest (ROI). The vertical portion of the T measured 335 mm × 50 mm and served as the site of fixation for the reference base. Eight PVC markers (2 mm in diameter) were placed on the ROI. The reference base was subsequently placed 100 mm from the isocenter of the ROI and secured to the vertical portion of the T-shaped specimen. The specimen was then placed on a custom-built carbon table, and scanning was performed with the Siremobil® Iso-C^3D (Figure 1).

Figure 1. A) Experimental setup for Iso-C^3D, showing the T-shaped specimen composed of synthetic bone material on a carbon table. The C-bow of the Iso-C^3D and the navigation system can be seen in the background. B) Experimental setup for 2D navigation: the pointer is fixed in a 3D holder for the reverse verification. [Color version available online.]

With the help of a laser reticule attached to the Siremobil® Iso-C^3D, the T-shaped specimen was placed in the isocenter of the scanner. The 3D data set obtained was then transferred to the navigation system in Iso-C^3D mode. The navigation system used was the SurgiGATE® system (Medivision, Oberdorf, Switzerland), which included an Ultra 10 Workstation (SUN Microsystems, Palo Alto, CA) and an Optotrak 3020 optoelectronic localizer (Northern Digital, Inc., Waterloo, Ontario, Canada). The precision of the optoelectronic localizer has been determined by the manufacturer to be 0.1 mm in translation and 0.1° in rotation. The localizer was placed 2 meters from the isocenter during each experiment. Altogether, 25 Iso-C^3D scans were obtained. Specifically, 5 scans with a stainless steel Schanz screw (Figure 2A), 5 with a titanium Schanz screw (Figure 2B), 5 with a 5-hole plate (Figure 3A), 5 with a PEEK Schanz screw (Figure 4), and 5 without any implant present.

Figure 2. A) Screenshot showing artifacts seen with the use of a stainless steel Schanz screw. The PVC markers can be seen clearly. B) A titanium Schanz screw results in fewer artifacts, and the PVC markers can be seen clearly. [Color version available online.]

The registration data was analyzed by “reverse verification” using a pointer mounted to a purpose-built manipulable 3D holder [10]. All of the above experiments were also repeated with fluoroscopy-based navigation (Figure 3B). Two projections (a.p. and lateral) were acquired with the navigation-enabled C-bow (Exposkop 8000®, Ziehm, Nuremberg, Germany). The images were imported to the system, and the precision of registration was similarly analyzed by reverse verification. The mean error was calculated using the mean value of the 8 markers from the 5 Iso-C^3D scans in each set for both Iso-C^3D and fluoroscopy-based imaging.

Figure 3. A) Screenshot showing artifacts seen with a 5-hole plate: in this case, the extremely heavy artifacts impair localization of the PVC markers. B) Screenshot of the fluoroscopy-based experimental setup: the PVC markers can be seen easily. The stainless steel Schanz screw does not influence the registration. [Color version available online.]

Figure 4. A) Schanz screw made of polyether-ether-ketone (PEEK). B) Iso-C^3D scan of the PEEK Schanz screw, with no artifacts visible. The screw is well defined from the synthetic bone.

Results

The mean error increased as more metal was placed in the scan area (the mean error with no implant was 0.04 mm, compared to 0.7 mm with a 5-hole plate). The artifacts caused by the titanium Schanz screw were less than those caused by the steel Schanz screw; however, this difference was not statistically significant (p = 0.05). There was no difference between the mean error with the PEEK Schanz screw and that for the specimen with no screw or plate present (0.05 mm versus 0.04 mm, p = 0.05). The precision of fluoroscopy-based navigation was not affected by the presence of metal in the area of acquisition, and there was no significant difference with regard to error between the different groups with fluoroscopy-based navigation (Figure 5).

Figure 5. Histogram showing the mean error of registration for different implants. While artifacts caused by the presence of metal have a significant influence on the accuracy of the Iso-C scan, the fluoroscopy-based navigation system presents comparable accuracy in all cases. [Color version available online.]

Discussion

Anatomical pair-point registration remains a critical step in navigation assisted surgery. Several potential sources of error have been cited in the literature, including incomplete data acquisition, improper segmentation, and bent or distorted instruments [10–12]. With increased experience, most of these sources of error can be minimized [13]. The use of self-acting registration during data acquisition also reduces the incidence of errors. Despite these precautions, metallic reference bases or implants placed too close to the isocenter can lead to extensive artifacts.

The results of this study demonstrate that the precision error of registration can be significantly impaired when metal is located within the scan area. In such cases, the mean error increased to values of between 0.55 and 0.7 mm, whereas the mean error with no implant and the PEEK screw amounted to 0.04 mm and 0.05 mm, respectively. While not commonly discussed as a potential source of error, this study illustrates the potential influence of metal artifacts when using Iso-C^3D navigation. The clinical relevance of this information may be of particular importance when considered in conjunction with other potential sources of error. While the clinical relevance of these factors has been found to be less critical when using navigation for spine or pelvic surgery [6], [14], an awareness of these potential sources of error will allow the surgeon to address and reduce these problems [13], [15]. Metallic artifacts were not found to significantly affect the precision of registration with fluoroscopy-based navigation.

Polyether-ether-ketone (PEEK OPTIMA®) is a key component in a number of implants [16–18] because of the several advantages it offers, including adaptability to sterilization, radiolucency, and superior mechanical properties. The use of PEEK-based instruments in our experiments resulted in improved precision of anatomic pair-point registration as compared to steel or titanium implants. The most significant downside to using PEEK OPTIMA® Schanz screws is the fact that no system is yet commercially available. The development of new reference clamps containing fewer metallic components and more radiolucent material (i.e., PEEK) appears to be of benefit in improving the precision of Iso-C^3D-based navigation. While metallic artifacts were not found to significantly affect precision with fluoroscopy-based navigation, they can nonetheless cause masks or shadows which can in turn make data acquisition more difficult. As such, the use of PEEK-based instruments may also prove advantageous in fluoroscopy-based navigation.

Conclusion

There are several potential sources of error in computer assisted surgery. They begin with the placement of the reference base in unfavorable anatomic positions, and continue with loosening during navigation, errors with hardware, excess play or bending of the drill, and errors in acquisition of data, registration, and verification. A standardized procedure is necessary to reduce these sources of error. Continued development and evolution of existing systems and instrumentation are essential to advance the future of navigation.