Image-free cup navigation inaccuracy: A two-study approach

Abstract

The present study tested the accuracy of an image-free navigation system used for total hip arthroplasty (THA). Two parallel, prospective studies were performed: one on real patients and the other on pelvic phantoms. We used a comparison between the intra-operative cup orientation, as displayed by the navigation system, and the post-operative cup position, as measured on CT data. The mean intrinsic overall error (± standard deviation) found in the phantom study was 2.6 ± 1.1° (range: 1.5-4.4°) for cup abduction, and 0.9 ± 0.7° (range: 0-2.5°) for cup anteversion. The system was less accurate in the clinical operative setting. The evaluation model was able to identify, and more importantly quantify, the clinically induced error. Ameliorating this would improve the clinical accuracy of the system.

• Accuracy
• image-free navigation
• total hip arthroplasty

Introduction

Consistently achieving correct alignment of implant components in total hip arthroplasty (THA) is one of the main challenges facing orthopaedic surgeons today. Navigation tools should play a major role, by offering three-dimensional information in real time concerning the tool positions and pelvic orientation during surgery. Hip navigation systems share some basic concepts regarding the reference system and angle definitions for implant alignment. For the acetabular cup, the reference system is based on the anterior pelvic plane (APP), defined by the two anterior iliac spines and the midpoint between the pubic tubercles. However, there is significant anatomic variation of the APP position in the population. This is one reason why the landmark-based navigation concept using the APP has not yet been clinically validated.

There is very little clinical data in the literature regarding system validation by accuracy measurement [1]. It is important that accuracy be tested for each system using standardized methods. As the system user, the clinician's role in elaborating these methods is evident. For this reason, the present study was dedicated to accuracy measurements of an image-free hip navigation system. The purpose was to measure the overall system accuracy and quantify the influence of the clinical application on this global error. The working hypothesis was that significant error could be introduced by anatomic landmark palpation during surgery. This error, induced by the clinical application, has not yet been measured and published in the literature.

Methods

The study reports accuracy measurements only for the acetabular component of the prosthesis. The cup alignment can be reported in different ways, using radiographic, anatomic or operative reference systems [2]. For this study, all measurements are reported in terms of cup anteversion and abduction using an anatomic reference, the APP. The present methodology is based on comparison between the intra-operative cup orientation, as displayed by the navigation system, and the post-operative cup position, as measured on CT data.

Two parallel prospective studies were done: one on real patients and the other on pelvic phantoms. The two studies used the same methodology, based on the comparison mentioned above regarding cup orientation. The clinical study was intended to determine the overall navigation accuracy by measuring the global error (E_g). The phantom study was performed using the most optimal setup for the navigation system, and was designed to measure the intrinsic system accuracy. It was intended to evaluate the intrinsic error (E_i) independent of user or clinical setting. The difference between the global and intrinsic error is termed clinically induced error (E_c). This indirect method was chosen because of the difficulty of directly measuring the E_c, considering the anatomic variability and the restrictive surgical environment.

The study also evaluated the HIPPILOT®, a software application dedicated exclusively to implanting the Symbios SPS Modular Neck hip prosthesis (SYMBIOS Orthopédie SA, Yverdon-les-bains, Switzerland). This software is designed for use with the Surgetics™ workstation (Praxim-Medivision, La Tronche, France).

Five Sawbones® pelvic phantoms were used. The navigated cup placement was performed on both sides, with a total of 10 cups being implanted. The anatomical landmarks were directly palpated, with no sterile draping being used. The final cup orientation was recorded for each operated side (Figure 1), and post-operative pelvic CT scans were performed to enable cup alignment measurements. The difference between post-operative and intra-operative cup orientations was determined in order to calculate the so-called intrinsic system error.

Figure 1. The phantom study. The cup was placed using a computer-guided cup tool and the final orientation was recorded by the system.

Five patients underwent navigated THA using the HIPPILOT® software. The same acetabular implants as above were used, and their final orientation was recorded by the system. All the procedures were performed by an experienced senior orthopaedic surgeon. As for the phantom study, post-operative CT scans were performed using the same protocol. The cup orientation was measured with specially designed software that used the same anatomic reference system as the navigation software. The global mean error in terms of anteversion and abduction was then calculated.

All phantoms and patients underwent post-operative CT-scan acquisition using the same protocol. A helical CT scanner was used (Somaton Sensation 16, Siemens AG, Munich, Germany) and provided 1.5-mm-thick slices of the entire pelvis (120 kV, 100-170 mAs).

An important aspect of the study was the development of specially designed software, named Cotyle Evaluator, to allow anatomic CT cup measurements (Figure 2). This software uses the following steps:

1. The CT data is uploaded as DICOM files and segmented.

2. The anatomic landmark points (anterosuperior iliac spines and the midpoint between the pubic tubercles) are selected using an intuitive interface. The reference system is then defined by the software, using the above points.

3. A CAD (computer aided design) cup model is displayed on the screen. The model is then superimposed on the actual cup and its position is manually modified until the best possible alignment is obtained. This position is defined as the post-operative cup anteversion and abduction.

4. Additionally, the software calculates and displays the pelvic flexion angle.

As the software is user-dependent, the intra- and inter-observer variability was determined. Five orthopaedic surgeons were asked to perform the same post-operative measurements on a pelvic phantom to calculate inter-observer variability. All measurements were performed 10 different times by one of the surgeons, allowing calculation of the mean intra-observer variability.

Figure 2. Screen view of the software interface used to determine the cup orientation on post-operative CT. The final position is displayed and recorded. [Color version available online.]

The cup position, orientation errors and pelvic flexion were reported in terms of mean, standard deviation and range using descriptive statistic tools. These values were positive or negative, depending on the calculated differences. Inter- and intra-observer variability was expressed by calculating the variance coefficient VARP. The absolute error values are reported as well. The statistical analysis included Student's t-test to compare means of two parametric variables. All tests were considered statistically significant if the probability coefficient p was less than 0.05.

Results

Intrinsic system accuracy

The mean intra-operative cup abduction angle (± standard deviation) was 48.9 ± 5.1° (range: 37.5-56.5°) and the mean intra-operative anteversion was 11.4 ± 5.8° (range: -2 to 18.5°). On post-operative CT, the mean cup abduction was 49 ± 6.8° (range: 35.2-59°), and the mean anterversion was 12 ± 6.5° (range: -1.7 to 22.1°). Pelvic flexion angle varied from -7.1° to 4.2°, with a mean of 0.1 ± 4.4°.

Measurement variance was expressed by calculating the variance coefficient (VARP). The mean intraobserver VARP for cup abduction was 0.4 ± 0.3 (range: 0.09 to 1), while the mean VARP for anteversion was 0.8 ± 0.5 (range: 0.1 to 2.1). The inter-observer VARP was 0.7 for cup abduction and 0.8 for anteversion.

The mean absolute intrinsic (E_ia) error for cup abduction was 2.6 ± 1.1° (range: 1.5-4.4°), while the mean absolute intrinsic error for anteversion (E_iv) was 0.9 ± 0.7° (range: 0-2.5°) (Figure 3).

Figure 3. Graphic representation of mean intrinsic error distribution for cup abduction and anteversion.

Overall and clinical accuracy

Five patients (4 male, 1 female) were included in the study. Their mean age was 51 years (range: 17 to 81 years). The mean intra-operative abduction angle was 46.6 ± 3° (range: 43-50°), and the mean intra-operative anteversion angle was 26.6 ± 14.3° (range: 13-39°). Post-operatively, the mean abduction was 49.2 ± 11.8° (range: 35.6-64.6°) and the mean anteversion was 18.4 ± 6.1° (range: 13.5-30.6°). The pelvic flexion varied from -18.4° to -5.3°, with a mean of -9.4 ± 6°.

The global mean error for abduction (E_ga) was 10.3 ± 5.6° (range: 2.1-16.7°) and the mean error for anteversion (E_gv) was 11.9 ± 8.9° (range: 0.7-25.3°).

The difference E_g - E_i was interpreted as the error due to the clinical application and was termed clinical error (E_c). The mean clinically induced error in cup abduction was 7.7°, and that in anterversion was 11°.

Discussion

Total hip arthroplasty is currently the major focus of navigation research teams [3]. Although different types of system are already in clinical use, significant controversy persists. The optimal implant position is not yet defined, and reference systems used to report results do not always correspond to the intra-operative reference, creating additional confusion. Moreover, the increased accuracy claims are not always sustained by reports in the literature. The few existing studies on system accuracy measurement used different methodologies, and no standardized clinical accuracy evaluation model can be found in the literature [1].

Validation of a new clinical accuracy measurement model

We have shown that the simple clinical end-to-end accuracy evaluation model proposed offers enough information to the surgeon. It is a two-step method, which first allows a determination of whether the system is sufficiently accurate in a “machine ideal” setting. If the intrinsic error is acceptable, the second step provides the end-to-end error, this time using a “real life” operative setting. The simple difference between the errors obtained during each step is logically the result of the clinical application. It is important to quantify this clinically induced error and identify its sources. There is no consensus in the literature concerning the level of clinical acceptability for these errors; however, the authors consider the intrinsic error to be acceptable in clinical terms for abduction and very weak for cup version.

All the post-operative CT scans were analyzed with the specially designed software, allowing real anatomic measurements [4]. The Cotyle Evaluator software was validated on pelvic phantom CT data, yielding very encouraging inter- and intra-observer variability scores. Its main advantage is the possibility of using the same anatomic reference system as the navigation software.

The clinical study group was small, but the results obtained are valuable for the system evaluation and its amelioration. The mean global abduction and anteversion errors were significant, and it is obvious that the “real life” operative setting reduced the overall system accuracy. Anatomic bone landmark palpation was done in lateral decubitus through soft tissues and sterile draping. To our knowledge, there are no other clinical studies in the literature reporting the global error and clinically induced error for an imageless hip navigation system.

Mor and coworkers [1] are the only authors who have described a similar methodology for hip navigation accuracy measurements. Their study, realized on two pelvic phantoms, obtained good end-to-end cup orientation accuracy for a CT-based hip navigation system.

Conclusion

Improving navigation tools is not possible without solid validation studies, especially concerning system accuracy. Clinicians should initiate and elaborate evaluation strategies in order to measure their confidence in these new tools. The present study developed and validated a simple model of system accuracy, and could eventually be applied to other types of navigation systems.