How Do Classic (Static) RSA and Patient Motion Artifacts Affect the Assessment of Migration of a TKA Tibial Component? An In Vitro Study : Clinical Orthopaedics and Related Research®

Secondary Logo

Journal Logo

BASIC RESEARCH

How Do Classic (Static) RSA and Patient Motion Artifacts Affect the Assessment of Migration of a TKA Tibial Component? An In Vitro Study

Cao, Han MD1,2; Sesselmann, Stefan MD3; Xu, Jing MD4; Seehaus, Frank PhD1,5; Forst, Raimund MD1

Author Information
Clinical Orthopaedics and Related Research 481(2):p 400-412, February 2023. | DOI: 10.1097/CORR.0000000000002453
  • Open

Abstract

Introduction

Aseptic loosening is one of the leading reasons for revision of TKA [22, 31]; one study suggested that 31% of all revision procedures are related to aseptic loosening [22]. Currently, the gold standard to assess implant-to-bone movements in vivo—referred to as migration—is classic Roentgen stereophotogrammetric analysis (RSA) [34, 37]. Unlike dynamic RSA that assesses joint kinematics [7, 29], classic RSA is a static measurement technique that is mainly used to assess the components of a joint implant [13, 38]. Classic RSA can predict an increased risk of subsequent implant revision after the first 2 years postoperatively, owing to a demonstrated correlation between early implant migration and the possibility of later aseptic loosening [9, 18, 27, 32]. In a recent study, RSA-tested implants (including 236 designs) had a lower revision rate at 10 years of follow-up than non-RSA-tested implants (including 103 designs; 4.4% versus 5.5%); thus, for all new designs of orthopaedic implants, testing with RSA is advocated [11]. In general, there are two approaches to RSA [19, 38]. Marker-based RSA uses markers that are physically attached to the implant being studied. On the other hand, with model-based RSA, virtual projections of a three-dimensional (3D) model using the contours of a radiographic projection of the implant are used instead of physical markers [16]. The latter has obvious appeal because implants do not need to be modified. Prior studies have suggested these two RSA approaches deliver comparable precision [15, 17, 33]. To avoid measurement errors because of nonsynchronized (within 0.3 seconds) [3] X-ray tube exposures, for RSA setups with nonelectric synchronization, it is important that a radiology technical assistant simultaneously pushes both X-ray tube exposure buttons [36]. Otherwise, patient motion artifacts may occur. An electronic synchronization device is recommended for RSA, if available, but this is not the current standard in radiology departments.

The effect of patient motion artifacts, to our knowledge, has not been tested for the marker-based or model-based RSA approaches, and this would be important to test, because even small amounts of movements might be important, given the high degree of precision we seek from RSA studies. Although the two kinds of RSA seem similar, neither has been tested in terms of how they perform regarding patient motion.

We therefore sought to assess (1) the effect of possible patient motion on the precision of RSA and (2) apparent differences in implant migration among axes (in-plane and out-of-plane translations and in-plane and out-of-plane rotations) of possible motion artifacts.

Materials and Methods

Experimental Overview

We performed a comparative RSA study (Fig. 1) using TKA components implanted in bone-implant models. Two such models were made so we could run our analyses in parallel, one with and one without simulated patient motion. We used one model to simulate patient motion in different directions and magnitudes, and kept the other model motionless as the baseline.

F1
Fig. 1:
This flow diagram shows an overview of the comparative RSA study.

Our primary study endpoint was the effect of patient motion on the precision of the estimation of RSA, which we measured through repeat examinations (“double examinations”, as termed in the RSA field) in a short time interval. Our secondary endpoint was to subanalyze the apparent differences among axes by evaluating in-plane and out-of-plane translations and rotations, which we did by comparing implant migration between two bone-implant models and between the marker-based and model-based RSA approaches.

Measurement Setup

RSA radiographic image pairs were acquired using one stationary tube (Multix RD 82477-01 Vertix ACS, Siemens) and one mobile X-ray tube (Mobilett Plus, Siemens) in a uniplanar RSA setup (Fig. 2A). Both X-ray tubes were arranged at an angle of 40° to each other and 140 cm over a calibration box (Umea Cage 43, RSABioMedical Innovations AB). Images were acquired at a setting of 81 kV and 5 mAs for each of the two X-ray tubes under the condition of nonelectronic synchronization.

F2
Fig. 2:
The experimental setup and phantom for RSA image pairs are shown. (A) Two radiography tubes were arranged with a phantom in the foci of two beams, with a uniplanar calibration box underneath. (B) The phantom consisted of two micromanipulators and two (simulated motion and no-motion) bone-implant models. (C) A pair of resulting RSA images is shown with three-dimensional (3D) reconstructed models of two tibial TKA components (the green implant is marked for the simulated motion protocol and the red one is marked for the no-motion protocol) and bone-implant markers. The orange, yellow, and green circles represent the indicator, fiducial, and control markers, respectively.

Phantom and Model

Experiments with additional radiation exposures of 1 millisievert [36] per radiograph for a patient who underwent TKA might be not approved as the RSA-tested object by a medical ethics committee. Therefore, a customized phantom (Fig. 2B) was developed to mimic the clinical situation of a patient who underwent TKA and is lying in the supine position. Two tibial tray models were established using proximal Sawbones tibiae (Sawbones: tibia with 12.5-mm Canal Solid Foam, Research Laboratories Inc) and two tibial components (BPK-S, Peter Brehm). Each proximal Sawbones tibia was injected with eight spherical 1-mm-diameter tantalum beads (Tantalum Markers 20402, Tilly Medical Products AB), following a standardized protocol [13, 38], to distribute the markers. Three additional tantalum beads were attached to each of the tibial components to simulate marker-based RSA. Then, tibial components were inserted in each proximal Sawbones tibia (Fig. 2B). Each of these bone-implant models was rigidly fixed to micromanipulators (Mitutoyo) on a base plate (Video 1; https://links.lww.com/CORR/A965).

Measurement and Analysis Protocol

The two bone-implant models in this phantom setup enabled us to simulate two measurement protocols, running in parallel, because both models could be shown in the same pair of acquired RSA images for a direct comparison of one pair (Fig. 2C), rather than comparison of two pairs. In the simulated motion protocol, a simulated motion bone-implant model (Fig. 2B) was used to simulate a defined patient movement (micromotion) resulting in motion artifacts. In the no-motion protocol, a no-motion bone-implant model (Fig. 2B) was used to simulate no patient movements (no motion artifacts); this no-motion protocol was the baseline for comparison with the simulated motion protocol.

The bone-implant models should only mimic patients’ possible motion (for motion artifacts), not possible migration between the bone and implant. The simulated motion bone-implant model (Fig. 2B) could be moved using micromanipulators with four degrees of freedom (DOF): three types of translational motion along the medial-lateral, cranial-caudal, and AP axes and one type of rotational motion around the cranial-caudal axis. The global coordinate system was defined relative to the calibration box. Translations along the medial-lateral (x) and cranial-caudal (y) axes constituted in-plane patient motion, and translations along the AP axis (z) were considered out-of-plane patient motion. For rotations around the posterior-anterior axis (Rz), simulated patient motion could be described as in-plane, and out-of-plane patient motion was rotation around the medial-lateral (Rx) and cranial-caudal axes (Ry).

Both protocols were established and followed in parallel for each series of RSA radiographs (Fig. 3). An RSA radiograph series consisted of a reference and follow-up pairs of radiographic images. Between the reference and the follow-up examinations of the RSA radiographic series, the no-motion and simulated motion protocols were performed in parallel for the four DOF. This simulation of possible patient triaxial motion of the knee was intended to mimic possible internal-external rotations of the legs of a patient who underwent TKA as well as muscle-twitching leg movements [8].

F3
Fig. 3:
The no-motion and simulated motion protocols were performed in parallel between the reference and follow-up examinations for three translational movements and one rotational movement. The no-motion protocol was the baseline for comparing migration. Expected implant migration was zero for both protocols.

The standard operating procedure for the no-motion protocol stated that there were no movements of the no-motion bone-implant model along and around the three axes (six DOF) in the global coordinate system. Meanwhile, in the simulated motion protocol, three translational (x, y, and z) movements and one rotational (Ry) movement of the simulated motion bone-implant model was performed in the following steps (Fig. 3). The intervals of the set-point value for the simulated motion protocol were 1 mm (range 0 to 5 mm; increment 1 mm) for each translational direction along the medial-lateral, cranial-caudal, and AP axes (motion direction after a muscle twitch), and 1° (range 0° to 5°; increment 1°) for rotations around the cranial-caudal axis (motion direction after an internal-external leg movement).

Each RSA radiographic (35 cm × 43 cm, 150 dots per inch, 2320 × 2828 pixels, and eight-bit gray-level resolution) series for each DOF was repeated five times. To compare migration resulting from both protocols, we considered the no-motion protocol as the baseline. Expected implant migration was zero, representing the baseline value for the no-motion and simulated motion protocols while no implant-to-bone motion was simulated. Migration was calculated using a model-based RSA software package (MBRSA 4.1, RSAcore). Implant migration was calculated as the standard outputs of three directions of translations and angular rotations (six DOF, named Tx, Ty, Tz, Rx, Ry, and Rz by the software) and the maximum total point motion (a translation vector of the point with the greatest combined motion). Migration resulting from the no-motion and simulated motion protocols could only be determined for simulated patient motion along the medial-lateral (x) and AP (z) axes (0 to 5 mm) and around the cranial-caudal (Ry) axis (0° to 5°). For translational motion along the cranial-caudal axis (y), we could calculate migration only for the no-motion protocol. It was impossible for us to calculate migration in the simulated motion protocol because there were missing 3D constructed beads or implants that were affected by motion artifacts. The software output described migration as -99.9.

Statistical Analysis

A condition number of ≤ 150 m-1 and a mean error of rigid body fitting of ≤ 0.35 mm were selected as cutoff values for validation analyses [32, 38]. Vector values of three translational axes (total translation) were calculated via the Pythagorean theorem (square root of x2 + y2 + z2). Precision (accuracy of zero motion) was determined through two examinations (within a time interval of 10 to 15 minutes) [13, 38] using the recommended formula: precision = ± (t) × SD, where “t” is the constant (2.262) obtained from the t-table, adjusted for the number of observations, and SD is the standard deviation calculated from 0 with the assumption there was no systemic bias [2]. Bland-Altman plots were created for a direct comparison by assessing the degree of agreement between the two different measurement methods [1]. The acceptable limits of agreement (mean difference ± 1.96 × SD) were defined as ± 0.5 mm for translations and ± 1.15° for rotations [12, 39]. The graphic approach showed any systematic differences between the two measurements and range of agreement, but it did not directly show whether agreement between the two methods was sufficient [23]. The SPSS statistical software package (version 24, IBM Corp) was used to calculate the mean and SD for migration values for each axis and the maximum total point motion. The bias of the maximum total point motion for model-based RSA was ± 0.12 mm (three to four times greater than that for marker-based RSA), and the precision of the maximum total point motion for model-based RSA ranged from 0.25 to 0.3 mm, or double that for marker-based RSA [26, 28, 40]. To test deviation of the migration result in case of simulated patient movement (motion artifacts), an analysis of variance was performed. The data were analyzed for normality using the Shapiro-Wilk test. A paired t-test and Wilcoxon test were used to determine p values, with p values < 0.05 were considered statistically significant.

Results

The Effect of Simulated Patient Motion on the Precision of RSA

The effect of simulated patient motion on the precision of RSA was generally negligible except for the motion direction of translations along the cranial-caudal axis (y), where it was large. However, the effect of simulated patient motion was greater in the model-based RSA approach than it was in the marker-based RSA approach. The precision of measured vectors of the RSA method, recorded as the baseline resulting from the no-motion protocol, was independent from the simulated patient movements. In marker-based RSA, for the simulated in-plane translations, that is, along the medial-lateral axis (x) (Table 1) and cranial-caudal axis (y) (Table 2), the mean ± SD migration values were between 0.016 ± 0.010 mm and 0.027 ± 0.011 mm, and between 0.015 ± 0.005 mm and 0.034 ± 0.011 mm for the out-of-plane translations (Table 3). For the in-plane rotations (Table 4), values were between 0.012 ± 0.005 mm and 0.028 ± 0.012 mm. In model-based RSA, values were between 0.025 ± 0.008 mm and 0.044 ± 0.020 mm, between 0.037 ± 0.009 mm and 0.057 ± 0.018 mm, and between 0.027 ± 0.010 mm and 0.129 ± 0.057 mm. The worst precision values were for simulated patient translations along the medial-lateral axis (x) (± 0.061 mm in model-based RSA and ± 0.025 mm in marker-based RSA), translations along the AP axis (z) (± 0.065 mm in model-based RSA and ± 0.029 mm in marker-based RSA), and rotations around the cranial-caudal axis (Ry) (± 0.128 mm in model-based RSA and ± 0.028 mm in marker-based RSA). The mean precision values in model-based RSA were 0.035 ± 0.015 mm, 0.045 ± 0.014 mm, and 0.049 ± 0.036 mm greater than those in marker-based RSA, in accordance with the simulated motion protocol in translations along the medial-lateral axis (x) (0.018 ± 0.004 mm; p = 0.01), translations along the AP axis (z) (0.018 ± 0.007 mm; p = 0.003), and rotations around the cranial-caudal axis (Ry) (0.017 ± 0.006 mm; p = 0.02).

Table 1. - Precision of calculated migration in accordance with the simulated motion protocol in translations along the medial-lateral axis (x)
Simulated motion (x) Model-based RSA in mm Marker-based RSA in mm
Precision Mean ± SD Precision Mean ± SD
Tx Ty Tz Vector Tx Ty Tz Vector
Reference ± 0.039 ± 0.036 ± 0.073 ± 0.031 0.040 ± 0.014 ± 0.019 ± 0.013 ± 0.027 ± 0.016 0.022 ± 0.007
0 mm ± 0.028 ± 0.011 ± 0.066 ± 0.032 0.029 ± 0.014 ± 0.016 ± 0.031 ± 0.024 ± 0.022 0.016 ± 0.010
1 mm ± 0.043 ± 0.040 ± 0.087 ± 0.046 0.044 ± 0.020 ± 0.024 ± 0.024 ± 0.040 ± 0.013 0.023 ± 0.006
2 mm ± 0.018 ± 0.012 ± 0.081 ± 0.061 0.038 ± 0.024 ± 0.016 ± 0.036 ± 0.029 ± 0.025 0.027 ± 0.011
3 mm ± 0.013 ± 0.035 ± 0.044 ± 0.018 0.025 ± 0.008 ± 0.013 ± 0.031 ± 0.037 ± 0.016 0.022 ± 0.007
4 mm ± 0.020 ± 0.031 ± 0.043 ± 0.021 0.025 ± 0.009 ± 0.020 ± 0.020 ± 0.043 ± 0.016 0.022 ± 0.007
5 mm ± 0.037 ± 0.029 ± 0.079 ± 0.038 0.038 ± 0.017 ± 0.019 ± 0.041 ± 0.031 ± 0.020 0.024 ± 0.009
Data are presented as the mean ± SD and precision of translations along the medial-lateral (x), cranial-caudal (y), and AP (z) axes (Tx, Ty, and Tz, respectively). The precision values are shown according to the recommended formula: precision = ± (t) × SD, where “t” is the constant (2.262) obtained from the t-table, adjusted for the number of observations, and SD is the standard deviation calculated from 0 with the assumption that there was no systemic bias [2].

Table 2. - Precision of calculated migration in accordance with the simulated motion protocol in translations along the cranial-caudal axis (y)
Simulated motion (y) Model-based RSA in mm Marker-based RSA in mm
Precision Mean ± SD Precision Mean ± SD
Tx Ty Tz Vector Tx Ty Tz Vector
Reference ± 0.029 ± 0.044 ± 0.069 ± 0.037 0.036 ± 0.017 ± 0.019 ± 0.013 ± 0.027 ± 0.017 0.023 ± 0.008
0 mm ± 0.034 ± 0.058 ± 0.054 ± 0.042 0.053 ± 0.019 ± 0.033 ± 0.021 ± 0.041 ± 0.015 0.026 ± 0.006
1 mm N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
2 mm N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
3 mm N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
4 mm N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
5 mm N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
Data are presented as the mean ± SD and precision of translations along the medial-lateral (x), cranial-caudal (y), and AP (z) axes (Tx, Ty, and Tz, respectively). N/A = not applicable because calculation was impossible. The precision values are shown according to the recommended formula: precision = ± (t) × SD, where “t” is the constant (2.262) obtained from the t-table, adjusted for the number of observations, and SD is the standard deviation calculated from 0 with the assumption that there was no systemic bias [2].

Table 3. - Precision of calculated migration in accordance with the simulated motion protocol in translations along the AP axis (z)
Simulated motion (z) Model-based RSA in mm Marker-based RSA in mm
Precision Mean ± SD Precision Mean ± SD
Tx Ty Tz Vector Tx Ty Tz Vector
Reference ± 0.019 ± 0.020 ± 0.135 ± 0.065 0.055 ± 0.029 ± 0.006 ± 0.019 ± 0.054 ± 0.029 0.024 ± 0.013
0 mm ± 0.019 ± 0.033 ± 0.093 ± 0.044 0.043 ± 0.019 ± 0.013 ± 0.022 ± 0.020 ± 0.012 0.015 ± 0.005
1 mm ± 0.037 ± 0.056 ± 0.034 ± 0.021 0.037 ± 0.009 ± 0.006 ± 0.025 ± 0.046 ± 0.020 0.023 ± 0.009
2 mm ± 0.020 ± 0.072 ± 0.030 ± 0.036 0.038 ± 0.016 ± 0.017 ± 0.042 ± 0.048 ± 0.025 0.034 ± 0.011
3 mm ± 0.027 ± 0.048 ± 0.100 ± 0.052 0.055 ± 0.023 ± 0.018 ± 0.047 ± 0.026 ± 0.014 0.027 ± 0.006
4 mm ± 0.080 ± 0.066 ± 0.071 ± 0.056 0.052 ± 0.025 ± 0.017 ± 0.028 ± 0.018 ± 0.013 0.018 ± 0.006
5 mm ± 0.078 ± 0.068 ± 0.079 ± 0.042 0.057 ± 0.018 ± 0.013 ± 0.031 ± 0.026 ± 0.013 0.019 ± 0.006
Data are presented as the mean ± SD and precision of translations along the medial-lateral (x), cranial-caudal (y), and AP (z) axes (Tx, Ty, and Tz, respectively). The precision values are shown according to the recommended formula: precision = ± (t) × SD, where “t” is the constant (2.262) obtained from the t-table, adjusted for the number of observations, and SD is the standard deviation calculated from 0 with the assumption that there was no systemic bias [2].

Table 4. - Precision of calculated migration in accordance with the simulated motion protocol in rotations around the cranial-caudal axis (Ry)
Simulated motion (Ry) Model-based RSA in mm Marker-based RSA in mm
Precision Mean ± SD Precision Mean ± SD
Tx Ty Tz Vector Tx Ty Tz Vector
Reference ± 0.020 ± 0.017 ± 0.094 ± 0.042 0.043 ± 0.019 ± 0.023 ± 0.005 ± 0.053 ± 0.028 0.028 ± 0.012
± 0.029 ± 0.013 ± 0.054 ± 0.023 0.027 ± 0.010 ± 0.014 ± 0.023 ± 0.013 ± 0.012 0.012 ± 0.005
± 0.030 ± 0.047 ± 0.048 ± 0.024 0.036 ± 0.011 ± 0.016 ± 0.023 ± 0.043 ± 0.016 0.022 ± 0.007
± 0.074 ± 0.059 ± 0.116 ± 0.046 0.073 ± 0.020 ± 0.021 ± 0.030 ± 0.029 ± 0.020 0.021 ± 0.009
± 0.041 ± 0.043 ± 0.071 ± 0.038 0.040 ± 0.017 ± 0.013 ± 0.009 ± 0.026 ± 0.012 0.014 ± 0.005
± 0.090 ± 0.032 ± 0.066 ± 0.041 0.053 ± 0.018 ± 0.009 ± 0.022 ± 0.041 ± 0.020 0.021 ± 0.009
± 0.227 ± 0.035 ± 0.200 ± 0.128 0.129 ± 0.057 ± 0.013 ± 0.014 ± 0.030 ± 0.013 0.018 ± 0.006
Data are presented as the mean ± SD and precision of translations along the medial-lateral (x), cranial-caudal (y), and AP (z) axes (Tx, Ty, and Tz, respectively). The precision values are shown according to the recommended formula: precision = ± (t) × SD, where “t” is the constant (2.262) obtained from the t-table, adjusted for the number of observations, and SD is the standard deviation calculated from 0 with the assumption that there was no systemic bias [2].

Apparent Differences in Implant Migration of Motion Artifacts

The migration assessment of the two tested implants differed and depended on the motion’s direction and magnitude. When simulated motion was the translation along the medial-lateral axis (x), the calculated migration (six DOF) of the two measurement protocols showed a high level of agreement based on Bland-Altman plots (Fig. 4). However, for translations along the AP axis (z) (Fig. 5), the upper limit of agreement for migration in model-based RSA was at the unacceptable level (Ry) (1.58° > 1.15°) (Fig. 5E). Regarding rotations around the cranial-caudal axis (Ry), the migration results (Tx, Ty, and Rz) were clinically acceptable and within the limits of agreement (mean difference ± 1.96 × SD): 0.03 ± 0.09 mm, 0.01 ± 0.04 mm, and -0.08° ± 0.12° in model-based RSA and 0.01 ± 0.03 mm, 0.05 ± 0.08 mm, and -0.11° ± 0.18° in marker-based RSA (Fig. 6). For simulated patient movements along the medial-lateral axis (x), increasing migration values of the maximum total point motion, depending on ROM, were determined for marker-based and model-based RSA approaches (Fig. 7A and Fig. 7B). The maximum migration value was 0.368 ± 0.032 mm (p = 0.01) in model-based RSA after a possible patient motion of 5 mm (Fig. 7A). For simulated patient movements along the AP axis (z), increasing migration values of the maximum total point motion (0.495 ± 0.214 mm; p = 0.02 and 0.836 ± 0.200 mm; p = 0.01) were found only in model-based RSA after possible patient motion of 4 and 5 mm (Fig. 7C), not in marker-based RSA (Fig. 7D). The most pronounced influence of nonsynchronized recorded RSA radiographs was for migration, represented as the maximum total point motion for both RSA approaches while internal-external rotations of a patient’s leg were simulated (Fig. 7E and Fig. 7F). The maximum total point motion increased from 0.038 ± 0.007 mm (for the no-motion protocol) to 1.684 ± 0.038 mm (for the simulated motion protocol) (p < 0.001) for marker-based RSA and from 0.101 ± 0.027 mm (for the no-motion protocol) to 1.973 ± 0.442 mm (for the simulated motion protocol) (p < 0.001) for model-based RSA, and was the worst-case scenario regarding patient motion artifacts.

F4
Fig. 4:
These Bland-Altman plots illustrate agreement between the simulated motion and no-motion bone-implant models regarding migration calculated by the two methods according to the protocols in translations along the medial-lateral axis (x). The six DOF are (A) Tx, (B) Ty, (C) Tz, (D) Rx, (E) Ry, and (F) Rz in model-based RSA and (G) Tx, (H) Ty, (I) Tz, (J) Rx, (K) Ry, and (L) Rz in marker-based RSA. The solid horizontal line represents the mean of all the differences plotted, and the two dashed lines represent the limits of agreement (mean ± 1.96 × SD). DOF = degrees of freedom.
F5
Fig. 5:
These Bland-Altman plots illustrate agreement between the simulated motion and no-motion bone-implant models regarding migration calculated by the two methods in accordance with the protocols in translations along the AP axis (z). The six DOF are (A) Tx, (B) Ty, (C) Tz, (D) Rx, (E) Ry (upper limit of agreement at the unacceptable level), and (F) Rz in model-based RSA and (G) Tx, (H) Ty, (I) Tz, (J) Rx, (K) Ry, and (L) Rz in marker-based RSA. The solid horizontal line represents the mean of all the differences plotted, and the two dashed lines represent the limits of agreement (mean ± 1.96 × SD). DOF = degrees of freedom.
F6
Fig. 6:
These Bland-Altman plots illustrate agreement between the simulated motion and no-motion bone-implant models regarding migration calculated by the two methods in accordance with the protocols in rotations around the cranial-caudal axis (Ry). The six DOF are (A) Tx, (B) Ty, (C) Tz, (D) Rx, (E) Ry, and (F) Rz in model-based RSA and (G) Tx, (H) Ty, (I) Tz, (J) Rx, (K) Ry, and (L) Rz in marker-based RSA. The solid horizontal line represents the mean of all the differences plotted, and the two dashed lines represent the limits of agreement (mean ± 1.96 × SD). DOF = degrees of freedom.
F7
Fig. 7:
Calculated migration outputs are shown as the maximum total point motion in accordance with the simulated motion and no-motion protocols in translations along the medial-lateral axis (x) for comparison in (A) model-based and (B) marker-based RSA, in accordance with the protocols in translations along the AP axis (z) for comparison in (C) model-based and (D) marker-based RSA, and in accordance with the protocols in rotations around the cranial-caudal axis (Ry) for comparison in (E) model-based and (F) marker-based RSA. The dashed lines are the reported precision (0.3 mm) [28] of the maximum total point motion.

Discussion

Because the classic RSA method allows for a high-level, precise, and accurate assessment, studies have recommended that all new implants have a phased introduction and should be tested by RSA. This could lead to safer and more effective patient care [11, 25, 27]. RSA is classified into marker-based and model-based approaches; however, the effect of patient motion has not been analyzed and quantified with implant assessments using either approach. This in vitro study focused on simulated patient motion and motion artifacts regarding controlling the quality of RSA images. We found that simulated patient motion affected implant migration in the RSA assessment, but not RSA precision via the current standard of radiology departments, and that motion artifacts could result in measurement errors in a clinical RSA study. Clinical RSA studies should evaluate precision (accuracy of zero motion), implant migration (six DOF), and the maximum total point motion in TKA. Clinical precision as a standardized output is assessed by repeat examinations of all patients, because each patient has a unique bone marker configuration, according to RSA guidelines [38] and the International Organization for Standardization standard [13]. Nonetheless, because it is difficult to acquire simultaneous exposures of two X-ray tubes manually, previous in vivo studies may not have paid attention to the effect of patient (joint) motion on the precision of RSA or assessed implant migration. Those assessments might have been less precise and accurate than they could be; in the worst case, those estimates could be misleading because of this problem.

Limitations

This study has several limitations. The phantom motion did not mimic real patient motion, but major motion directions were included in this phantom study. The tibial component could not be rotated around all three axes because the customized phantom was equipped with only one spring-loaded microrotator (Ry) per micromanipulator. Other types of rotations (Rx and Rz) were not the major motion of the leg and thus were not considered necessary to evaluate. The geometry of the tibial implants was almost symmetrical along the cranial-caudal axis, and this factor might have affected the actual results of testing motion artifacts in the simulated motion protocol in rotations around the cranial-caudal axis (Ry). All six DOF of translations and rotations should be reported according to RSA guidelines [38]. Gudnason et al. [10] reported that rotations of a tibial component around the medial-lateral axis (Rx) demonstrated a strong association between tibial component micromotion measured after 2 years and the risk of revision for aseptic loosening up to 15 years. Gudnason et al. [10] also reported that rotations were slightly better at predicting aseptic loosening than the maximum total point motion through a comparison of the area under the receiver operating characteristic curves (80% versus 68%). Because rotations of the tibial component have not been reported in a widely agreed-upon manner, meaningful analyses for comparison with other scientific studies were not possible [30]. However, we reported the migration of six DOF. Only five series of phantom motions were evaluated in the statistical analysis. To our knowledge, the amount of expected patient motion via the current standard of radiology departments has not been reported; thus, we performed a preliminary experiment of the greatest amount (approximately 4 mm) of patient motion permitted in order to possibly calculate migration of the investigated implant. In the current study, intervals of phantom motion were 1 mm for translations and 1° for rotations. A smaller interval (such as 0.01 mm, 0.1 mm, or 0.5 mm) or a logarithmic progression of simulated micromotion might have been better for more detailed (smaller) thresholds than 1 mm or 1° in this study.

Another major limitation was that only one type of cage (a uniplanar calibration cage) was implemented in this phantom study. Other types were not assessed. A biplanar calibration cage is widely used in RSA studies of knee implants; however, it was unavailable in our laboratory and its use might have influenced the results of this study owing to its different distribution of fiducial and control markers. Because of this limitation, our study might have underestimated the effect of movements on the precision of RSA. Further studies should focus on the influence of different cages. In this study, the uniplanar calibration cage had 13 fiducial and 19 control markers for the 3D coordinate system, and because of the 19 control markers, there might have been differences between the simulated motion protocol in translations along the medial-lateral axis (x) and cranial-caudal axis (y) regarding the influence of simulated patient motion on the assessment of implant migration. Likewise, this study did not use two other special cages, either a universal calibration cage [4] or a thinner 250-mm uniplanar calibration cage with 40 fiducial and 30 control markers, which had superior precision for translations (SD ≤ 0.054 mm) in a phantom study [14] to that of the uniplanar calibration cage (SD ≤ 0.093 mm [14]) used in the current study. Some beads inserted into the models or markers in the calibration cage overlapped with the cage’s metal strip or the models. Because the number of markers in the cage is related to the accuracy of the 3D reconstruction [6, 41], an increase in the number of markers would increase accuracy but would only result in minimal improvement after 32 markers [4]. Future studies should increase the number of markers and attempt to avoid missing or occluded markers on radiographs.

The Effect of Simulated Patient Motion on the Precision of RSA

Motion did not have a large effect on the precision of RSA, except for the cranial-caudal axis (y), and model-based RSA and marker-based RSA differed. Model-based RSA was less precise but more robust than marker-based RSA regarding patient motion. We normally validate clinical RSA systems in a standardized manner by presenting precision values, which are assessed by repeat examinations [13, 38], but our important finding is that precision (repeat examinations) may not be more reliable than using the current standards that are widely used in radiology departments. When we acquire precision values for validation and comparison of RSA systems, they cannot inform us whether there is patient motion during RSA examinations. The precision of RSA varies between joints treated with arthroplasty and patients treated for spine conditions [5, 21, 35]; the translation varies between 0.15 and 0.60 mm and rotation varies between 0.3° and 2°, with a 99% confidence interval [20]. The clinical precision of translations in lower-extremity arthroplasty was reported as 0.06 mm for marker-based RSA and 0.11 mm for model-based RSA [15, 24]. In the current phantom study, the precision values were below the reported threshold of ± 0.3 mm for translations [2]; however, compared with the aforementioned higher threshold (0.11 mm) in model-based RSA, the only inferior value for precision was detected in phantom motion of 5° of rotations around the cranial-caudal axis (0.128 mm). If clinical studies achieve acceptable precision, it does not mean that patient motion did not occur during Roentgen-based image acquisition, because this type of patient motion (micromotion) cannot be identified by the naked eye. Two paired X-ray tubes for RSA (not the electronically synchronized X-ray setup) are normally exposed for radiograph acquisition by means of firing two ignition buttons simultaneously. However, this process is not entirely simultaneous because of manmade exposures. During the exposure of two paired tubes, there is still a short time difference (within 0.3 seconds) [3]. Therefore, it is difficult to ensure the investigated implants are absolutely still during radiograph acquisition. In clinical RSA studies, researchers and radiographers should pay special attention to the effect of possible patient motion on data (implant migration) collection, even if acceptable precision is achieved, as in the current study.

Apparent Differences in Implant Migration of Motion Artifacts

Simulated patient motion exceeding 1 mm or 1° on nonsynchronized RSA images affected measurement errors regarding migration of a tibial component. Motion artifacts are more susceptible to migration in model-based RSA than in marker-based RSA. Translations of the simulated motion bone-implant model along the cranial-caudal axis (y) were more sensitive than that along other axes regarding the detection of motion artifacts. When simulated motion was the translation along the AP axis (z), it was difficult to detect motion artifacts. This type of patient motion did not affect measurement errors. In all six DOF of migration, rotational migration could give more information (greater limits of agreement and more scatter on Bland-Altman plots) than translational migration. We found that the comparative assessment of six DOF of migration between the simulated motion implant and the no-motion implant was better than that of precision for detecting motion artifacts. The maximum total point motion indicates whether the in vivo implant is considered loose, and is the most-reported parameter in TKA [30]. The acceptable threshold of the maximum total point motion is 0.54 mm at 12 or 6 months postoperatively [30, 31], and was more recently modified to be between 0.57 and 0.64 mm [26]. The reported precision of the maximum total point motion was ± 0.3 mm [28]. Nevertheless, a direct comparison of the maximum total point motion between the two bone-implant models (simulated motion and no-motion) provided more-valuable information than a comparison between the model’s value (simulated motion) and the reported threshold; for example, we could identify that the 3-mm, 4-mm, and 5-mm set-point groups in the simulated motion protocol were different between the same groups in the no-motion protocol through our comparison of the two models, but we could not identify this through a comparison with the threshold (Fig. 7B). Therefore, among the standard reported parameters (precision, six DOF of implant migration, and maximum total point motion), the maximum total point motion is the most susceptible to patient motion and is the best parameter for monitoring possible patient motion. In clinical RSA examinations, we suggest that, in addition to repeat (double) examinations, a comparison of the maximum total point motion between two (double) implants, that is, between the tested (simulated motion) implant and the baseline (no-motion) implant, as in this study, is needed for data reliability because patient motion artifacts can be detected accurately with the maximum total point motion.

Conclusion

Patient motion artifacts affect the measurement accuracy of RSA. Precision (repeat examinations) may not be reliable via the current standard of radiology departments. Even if researchers achieve acceptable precision, in clinical RSA studies, there should be concern about the effect of patient motion on the reliability of data (implant migration). Specially trained radiographers are crucial in order to correctly acquire radiographs, especially when simultaneous X-ray exposures are not electronically automated. In general, RSA requires synchronized image acquisition, and this should become the state-of-the-art in RSA research. Considering that the maximum total point motion is the most susceptible to patient motion and is the best parameter for monitoring possible patient motion, in clinical RSA examinations, we suggest double examinations with double implants for RSA validation; that is, in addition to repeat examinations, the maximum total point motion should be compared between the tested implant and baseline (no-motion) implant because of the accurate detection of patient motion artifacts.

Acknowledgments

We thank the staff of the Department of Radiology in Malteser Waldkrankenhaus St. Marien, Erlangen, Germany, for their support. The present work was performed in partial fulfilment of the requirements for obtaining the degree of Dr. med. (HC) at Friedrich-Alexander-Universität Erlangen- Nürnberg (FAU), Erlangen, Germany.

References

1. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307-310.
2. Bojan AJ, Bragdon C, Jönsson A, Ekholm C, Kärrholm J. Three-dimensional bone-implant movements in trochanteric hip fractures: precision and accuracy of radiostereometric analysis in a phantom model. J Orthop Res. 2015:33:705-711.
3. Bojan AJ, Jönsson A, Granhed H, Ekholm C, Kärrholm J. Trochanteric fracture-implant motion during healing - a radiostereometry (RSA) study. Injury. 2018;49:673-679.
4. Broberg JS, Yuan X, Teeter MG. Radiostereometric analysis using clinical radiographic views: development of a universal calibration object. J Biomech. 2018;73:238-242.
5. Brodén C, Sandberg O, Olivecrona H, Emery R, Sköldenberg O. Precision of CT-based micromotion analysis is comparable to radiostereometry for early migration measurements in cemented acetabular cups. Acta Orthop. 2021;92:419-423.
6. Cai R, Yuan X, Rorabeck C, Bourne RB, Holdsworth DW. Development of an RSA calibration system with improved accuracy and precision. J Biomech. 2008;41:907-911.
7. Christensen R, Petersen ET, Jurgens-Lahnstein J, et al. Assessment of knee kinematics with dynamic radiostereometry: validation of an automated model-based method of analysis using bone models. J Orthop Res. 2021;39:597-608.
8. Crockarell RJ, Guyton LJ. Arthroplasty of the knee. In: Canale ST, Beaty HJ, eds. Campbell's Operative Orthopaedics. 11th ed. Mosby Elsevier; 2007:246-255.
9. Grewal R, Rimmer MG, Freeman MA. Early migration of prostheses related to long-term survivorship. Comparison of tibial components in knee replacement. J Bone Joint Surg Br. 1992;74:239-242.
10. Gudnason A, Adalberth G, Nilsson KG, Hailer NP. Tibial component rotation around the transverse axis measured by radiostereometry predicts aseptic loosening better than maximal total point motion. Acta Orthop. 2017;88:282-287.
11. Hasan S, Marang-van de Mheen PJ, Kaptein BL, Nelissen R, Pijls BG. RSA-tested TKA implants on average have lower mean 10-year revision rates than non-RSA-tested designs. Clin Orthop Relat Res. 2020;478:1232-1241.
12. Hurschler C, Seehaus F, Emmerich J, Kaptein BL, Windhagen H. Accuracy of model-based RSA contour reduction in a typical clinical application. Clin Orthop Relat Res. 2008;466:1978-1986.
13. ISO16087:2013(E). Implants for surgery — Roentgen stereophotogrammetric analysis for the assessment of migration of orthopaedic implants. International Organization for Standardization, Switzerland; 2013.
14. Jürgens-Lahnstein JH, Petersen ET, Laursen M, et al. Development, construction, and validation of a thinner uniplanar calibration cage for radiostereometry. J Orthop Res. 2022;40:1645-1663.
15. Kaptein BL, Valstar ER, Stoel BC, Reiber HC, Nelissen RG. Clinical validation of model-based RSA for a total knee prosthesis. Clin Orthop Relat Res. 2007;464:205-209.
16. Kaptein BL, Valstar ER, Stoel BC, Rozing PM, Reiber JH. A new model-based RSA method validated using CAD models and models from reversed engineering. J Biomech. 2003;36:873-882.
17. Kaptein BL, Valstar ER, Stoel BC, Rozing PM, Reiber JH. A new type of model-based Roentgen stereophotogrammetric analysis for solving the occluded marker problem. J Biomech. 2005;38:2330-2334.
18. Kärrholm J, Borssen B, Lowenhielm G, Snorrason F. Does early micromotion of femoral stem prostheses matter? 4-7-year stereoradiographic follow-up of 84 cemented prostheses. J Bone Joint Surg Br. 1994;76:912-917.
19. Kärrholm J, Gill RH, Valstar ER. The history and future of radiostereometric analysis. Clin Orthop Relat Res. 2006;448:10-21.
20. Kärrholm J, Herberts P, Hultmark P, Malchau H, Nivbrant B, Thanner J. Radiostereometry of hip prostheses. Review of methodology and clinical results. Clin Orthop Relat Res. 1997;344:94-110.
21. Keller MC, Hurschler C, Schwarze M. Experimental evaluation of precision and accuracy of RSA in the lumbar spine. Eur Spine J. 2021;30:2060-2068.
22. Lombardi AV Jr, Berend KR, Adams JB. Why knee replacements fail in 2013: patient, surgeon, or implant? Bone Joint J. 2014;96:101-104.
23. Nazari-Farsani S, Finnilä S, Moritz N, Mattila K, Alm JJ, Aro HT. Is model-based radiostereometric analysis suitable for clinical trials of a cementless tapered wedge femoral stem? Clin Orthop Relat Res. 2016;474:2246-2253.
24. Nazari-Farsani S, Vuopio M, Löyttyniemi E, Aro HT. Contributing factors to the initial femoral stem migration in cementless total hip arthroplasty of postmenopausal women. J Biomech. 2021;117:110262.
25. Nelissen RG, Pijls BG, Karrholm J, Malchau H, Nieuwenhuijse MJ, Valstar ER. RSA and registries: the quest for phased introduction of new implants. J Bone Joint Surg Am. 2011;93:62-65.
26. Niesen AE, Hull ML. Propagation of registration error into maximum total point motion to analyze tibial baseplate stability at six months using marker-based and model-based RSA. J Biomech. 2021;127:110651.
27. Nilsson KG, Kärrholm J. RSA in the assessment of aseptic loosening. J Bone Joint Surg Br. 1996;78:1-3.
28. Øhrn FD, Van Leeuwen J, Tsukanaka M, Röhrl SM. A 2-year RSA study of the Vanguard CR total knee system: a randomized controlled trial comparing patient-specific positioning guides with conventional technique. Acta Orthop. 2018:89:418-424.
29. Petersen ET, Rytter S, Koppens D, et al. Patients with knee osteoarthritis can be divided into subgroups based on tibiofemoral joint kinematics of gait - an exploratory and dynamic radiostereometric study. Osteoarthritis Cartilage. 2022;30:249-259.
30. Pijls BG, Plevier J, Nelissen R. RSA migration of total knee replacements. Acta Orthop. 2018;89:320-328.
31. Pijls BG, Valstar ER, Nouta KA, et al. Early migration of tibial components is associated with late revision: a systematic review and meta-analysis of 21,000 knee arthroplasties. Acta Orthop. 2012;83:614-624.
32. Ryd L, Albrektsson BEJ, Carlsson L, et al. Roentgen stereophotogrammetric analysis as a predictor of mechanical loosening of knee prostheses. J Bone Joint Surg Br. 1995;77:377-383.
33. Seehaus F, Olender GD, Kaptein BL, Ostermeier S, Hurschler C. Markerless Roentgen stereophotogrammetric analysis for in vivo implant migration measurement using three dimensional surface models to represent bone. J Biomech. 2012;45:1540-1545.
34. Sundfeldt M, Carlsson LV, Johansson CB, Thomsen P, Gretzer C. Aseptic loosening, not only a question of wear: a review of different theories. Acta Orthop. 2006;77:177-197.
35. Ten Brinke B, Beumer A, Koenraadt K, Eygendaal D, Kraan GA, Mathijssen N. The accuracy and precision of radiostereometric analysis in upper limb arthroplasty. Acta Orthop. 2017;88:320-325.
36. Valstar ER. RSA in practical steps. In: Digital RSA Development, Validation, and Clinical Application. Leiden University Press; 2001.
37. Valstar ER, Gill RH. Radiostereometric analysis in orthopaedic surgery: editorial comment. Clin Orthop Relat Res. 2006;448:2.
38. Valstar ER, Gill RH, Ryd L, Flivik G, Börlin N, Kärrholm J. Guidelines for standardization of radiostereometry (RSA) of implants. Acta Orthop. 2005;76:563-572.
39. Valstar ER, Nelissen RGHH, Reiber JHC, Rozing PM. The use of Roentgen stereophotogrammetry to study micromotion of orthopaedic implants. ISPRS Journal of Photogrammetry and Remote Sensing. 2002;56:376-389.
40. van Hamersveld KT, Marang-van de Mheen PJ, Nelissen RGHH. The effect of coronal alignment on tibial component migration following total knee arthroplasty: a cohort study with long-term radiostereometric analysis results. J Bone Joint Surg Am. 2019;101:1203-1212.
41. Yuan X, Ryd L. Accuracy analysis for RSA: a computer simulation study on 3D marker reconstruction. J Biomech. 2000;33:493-498.

Supplemental Digital Content

Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the Association of Bone and Joint Surgeons