iCT Conference 2014 – www.3dct.at 209

Investigation on the accuracy of CT measurements for wear testing of prosthetic joint components

Simone Carmignato1, Manuel Balcon2, Filippo Zanini1 1University of Padova, Department of Management and Engineering, Stradella San Nicola 3, Vicenza,

Italy, e-mail: simone.carmignato@unipd.it 2University of Padova, Department of Industrial Engineering, Via Venezia 1, Padova, Italy,

e-mail: manuel.balcon@unipd.it

Abstract Wear testing of prosthetic joint components is fundamental for understanding wear mechanisms and improving the quality of manufactured orthopaedic prostheses. To this end, accurate measuring solutions are needed for quantifying wear volumes and characterizing the geometry of worn bearing surfaces. In this work, capability and accuracy of industrial computed tomography in wear assessment of prosthetic components are investigates. Advantages and limitations are determined in comparison with measurements by coordinate measuring machines.

Keywords: Dimensional metrology, X-ray Computed Tomography, Joint prosthesis, Wear testing, Coordinate measurements, Volume measurements.

1 Introduction Wear and failure of prosthetic joint components is a major problem in orthopaedic joint replacement. For instance, in hip joint replacement, although replaced hip joints are currently expected to last 15–20 years, up to 10% of procedures carried out are subject to revision operations for replacing joints which have failed prematurely [1,2]. Wear volumes measurement and geometrical characterization of worn bearing surfaces are fundamental for understanding wear mechanisms and improving technological solutions provided by manufacturers of prosthetic joint components. Wear testing of prostheses is typically performed using wear testing machines that physically reproduces controlled loading and relative motion for millions of cycles on the prosthetic components to be evaluated [3]. The current standard method for evaluating wear volumes of prosthetic components is the gravimetric method, which has been proven to be sufficiently accurate [3]. This method uses a microbalance to measure the specimen weight, before and after wear. Then the weight loss is calculated as the difference of the two measurements. However, this method can quantify only globally the worn material, while it cannot provide local information on the worn surface geometry. In addition, it can be applied only for in-vitro testing of new prosthetic components and not for assessing retrieved parts (which may also contain residuals, such as adhesive or bone). Furthermore, other damages not involving material loss, such as plastic deformations, are not measurable by this method. Coordinate Measuring Machines (CMMs) with tactile probes have recently been used as an alternative to the gravimetric method [4, 5]. CMMs can be used to evaluate both wear volume and wear distribution over the worn surface. However, the uncertainty of wear volume measurements from CMM data is in some cases too high, especially when polymeric components are measured, due to several limitations, including: probing force, fixation method, number and distribution of measured points, and alignment procedures [6]. The present work investigates the capabilities and accuracy of X-ray Computed Tomography (CT) in wear assessment of prosthetic components. The paper includes uncertainty assessment for wear

210

volume measurement by CT. The results of wear volume and wear distribution measurements are compared with corresponding results obtained by CMMs on the same samples.

2 Materials and methods In this work, a number of samples have been tested and measured using different measuring systems, as described b.

2.1 Prosthetic joint components Both hip and knee prosthesis were studied in this work. Concerning hip prosthesis, two types of bearings were examined: ceramic-on-ceramic hip prosthesis (where both the femoral head and the acetabular component are of ceramic material, see Figure 1), and ultra-high molecular weight polyethylene (UHMWPE) cups (see Figure 2). Concerning knee prosthesis, two types of polyethylene unicondylar knee prostheses (UKP) were examined: UHMWPE tibial inserts with congruous design (see Figure 3-left) and UHMWPE tibial inserts with non-congruous design (see Figure 3-right).

Figure 1: Ceramic hip joint components: 32 mm femoral head (left), 28 mm femoral head (centre) and 32 mm

acetabular cup (right).

Figure 2: Polyethylene acetabular cups with 30 mm diameter of spherical bearing surface.

Figure 3: Polyethylene uni-condylar knee prostheses: six components with congruous design (left) and four

components with non-congruous design (right).

iCT Conference 2014 – www.3dct.at 211

In addition to new components, which were measured both in unworn and worn status (before and after controlled wear performed by wear testing machines), also retrieved components were measured, which were removed from patients after years of implantation; namely: a retrieved ceramic femoral head and a retrieved UHMWPE tibial insert with non-congruous design.

2.2 CT system CT measurements in this work were performed at University of Padova using an industrial CT system: Nikon Metrology MCT225 scanner, with 225 kV liquid cooled micro-focus reflection source and air-cooled cabinet. The maximum permissible error (MPE) for length measurements, stated by the manufacturer based on tests in accordance to the guideline VDI/VDE 2630 part 13, is MPE = 9 + L/50 μm (where L is the length in mm).

2.3 Coordinate measuring machines The specimens were measured also by CMMs, for comparison with CT measurements and for dimensional calibrations, in addition to gravimetric calibrations from a microbalance. Two CMMs were used: (i) ZEISS Prismo VAST 7 equipped with tactile scanning probe head and (ii) Werth Video-Check IP 400 multisensor CMM equipped with touch-trigger probe, video imaging probe and laser distance sensor based on the Foucault principle (knife edge). In this work, the first CMM was preferred for performing tactile measurements, because of its scanning head that allows faster probing (which is advantageous for ensuring sufficiently close-meshed distribution of measured points [7]). The second CMM was used for optical measurements; limitations of such optical measurements are shortly discussed in the following Sections 3.2 and 4.

2.4 Uncertainty determination method The uncertainty of volume measurements from CT data was determined applying the task-specific procedure proposed in [7], which consists of two steps: (i) identification of the most important uncertainty components, and (ii) overall uncertainty evaluation through experimental method using calibrated samples. The first step is aimed at quantifying the most significant error sources for subsequent optimization of measurement parameters; while the second step directly determines the overall measurement uncertainty using the substitution method derived from ISO 15530-3 [8]. This method was practically applicable in the cases when a calibrated object similar to the prosthetic components was available with sufficiently reduced calibration uncertainty (e.g. calibrated ceramic sphere in the case of ceramic femoral heads), but showed its limitations when similar calibrated objects with sufficiently reduced calibration uncertainty were not available (e.g. polymeric samples with more complex geometry). In these last cases, in fact, calibration by CMM measurements produced too large uncertainties and hence could not be used as valid calibration uncertainties without yielding large overestimation of CT measurement uncertainty according to ISO 15530-3. Therefore, in these cases other methods had to be used for uncertainty assessment. One method that produced reasonable uncertainty estimations was adapted from the approach presented in [9]. According to this method, the uncertainty of CT volume measurement is determined based on the consideration that the volume obtained by a CT measurement is the result of the product of the number of voxels (which is not necessarily an integer) and the voxel volume : ∙ ∙ , where is the voxel size. Then, the combined standard uncertainty on the measured volume can be calculated as:

∙ ∙ 2 ∙ ∙ ∙ , (1) where and are respectively the uncertainties of and , and , is the covariance. If voxel size determination and edge detection can be assumed uncorrelated, Eq. (1) can be reduced to:

212

∙ ∙ 3 ∙ ∙ ∙ ∙ (2) According to Eq. (2) and [9], this method assumes that the uncertainty on the measured volume consists of two terms, caused by respectively the uncertainty on the voxel size and the uncertainty on the edge detection.

3 Results In this section, the main results obtained from CT measurements are presented and subsequently compared with analogous results obtained from CMM measurements on the same samples.

3.1 Results of CT measurements The samples described in Section 2.1 were measured using the CT system cited in Section 2.2. VG Studio Max 2.2 software was used for surface determination on CT data, using standard adaptive threshold method, and for volume calculation of the measured prosthetic components. In the following, examples of results are shown for ceramic and polymeric components. Figure 4 shows the wear map for a ceramic femoral head with nominal diameter of 32 mm. The colour map shows the deviations, in millimetres, between the surface determined from CT measurement of the worn head and the surface of the unworn geometry. Gaussian best-fitting on unworn surface areas was used for the alignment. Volume calculation was performed using two different procedures. (1) The first procedure calculates the volume of the “top hemisphere” by computing the volume included between the measured surface and a plane through the centre of the sphere datum aligned to the unworn areas of the measured surface, according to the approach described in [6]. (2) The second procedure, instead, calculates directly the volume of the entire prosthetic components using the dedicated algorithm implemented in VG Studio Max 2.2. The expanded uncertainty (k=2) of volume measurements obtained using the first procedure were determined in a range between 25 and 35 mm3 (which correspond to 0.20-0.28% of the sample volume), depending on the extension of the worn area on different worn heads. The second procedure allowed obtaining lower measurement uncertainties, below 20 mm3 (which correspond to 0.16% of the sample volume). The first procedure, in fact, is affected by additional errors introduced by the alignment to the reference datum, which are not present in the second procedure. The second procedure, hence, is always preferred when applicable (e.g. for CT measurement of non-retrieved components); therefore, all the uncertainties of volume measurements reported in this sub-section refer to the application of the second procedure. The second procedure, however, is not applicable for wear volume determination (difference of volumes before and after wear) in the case of retrieved components, where the original unworn volume is not known; in this case, the first procedure can be used assuming that the original unworn geometry corresponds to the datum aligned to the unworn areas of the measured surface.

Figure 4: Deviation map showing wear distribution for a ceramic femoral head with diameter of 32 mm and measured by CT system [7].

iCT Conference 2014 – www.3dct.at 213

Figure 5 shows the wear map of the spherical bearing surface (with nominal diameter of 30 mm) of an UHMWPE acetabular cup. The expanded uncertainty of volume measurement obtained in this case was below 30 mm3, which correspond to 0.15% of the sample volume. It is emphasized here that Feldkamp artifacts [10] may affect considerably the results visible in the wear maps obtained from CT measurements, if such artifacts occur in zones that are relevant for wear quantification. For this reason, the sample orientation during CT scanning should always be chosen carefully, so that Feldkamp artifacts do not appear on the relevant zones for wear visualization. Figure 6 shows the Feldkamp artifact obtained for the same component shown in Figure 5; the artifact is located in a zone that is not relevant for wear visualization. Feldkamp artifact appears typically also on other samples of different shapes, including femoral heads as the one shown in Figure 4; this artifact is not visible in Figure 4 because the sample was oriented so that the artifact was not appearing in the relevant wear zones. The errors introduced by the Feldkam effect are already taken into account in the volume measurement uncertainty.

Figure 5: Deviation map showing wear distribution on

the spherical bearing surface of a polyethylene acetabular cup measured by CT system.

Figure 6: Deviation map of the same component as in in Figure 5, showing the other side of the acetabular cup; a Feldkamp artifact is visible in the red zone.

Figure 7 shows the wear map of an UHMWPE non-congruous uni-condylar knee prosthetic component (see Figure 3-right). The expanded uncertainty of volume measurement obtained in this case was below 0.20% of the sample volume.

Figure 7: Deviation map showing wear distribution for a polyethylene non-congruous uni-condylar knee prosthetic

component measured by CT system.

3.2 Comparison with CMM results All prosthetic components measured by the CT system and shown in Section 3.1 were also measured using the tactile CMM described in Section 2.3, for comparing the measurement results. Contrary to CT, CMM measurements cannot scan the entire surface of the component in one measurement run, due to accessibility restrictions and to the fixture system used for clamping the

214

sample to the CMM table. Therefore, only the relevant areas of the surface are measured by the CMM for determining the wear distribution. This implies that the procedure used for volume calculation cannot calculate directly the entire volume of the prosthetic component from the measured surface (see procedure (2) in Section 3.1), but needs to calculate the volume included between the measured surface and a datum aligned to the unworn areas of the measured surface, according to the approach described in [6] (see procedure (1) in Section 3.1). In all the CMM measurement cases shown in the following, therefore, this procedure involving the alignment to a datum is always used. Figure 8-left shows the wear map obtained from tactile CMM measurements of the ceramic femoral head whose CT measurements were already presented in Section 3.1 (see Figure 4). Figures 4 and 8 compares the deviation maps for the cases of CT and CMM respectively. The expanded uncertainty of volume measurement obtained through tactile CMM in this case was 0.02% of the sample volume. This confirms that, in the case of ceramic prosthetic components, the tactile CMM can produce more accurate volume measurements than the CT system. This was already proven in [7]. Further validation of the CMM volumetric measurement method was performed also through comparison with gravimetric measurements, for nine different femoral heads [6].

Figure 8: Deviation maps showing wear distribution of a ceramic femoral head with diameter of 32 mm (same

sample as in Figure 4), measured respectively by: tactile CMM (left) and optical CMM (right) [7].

Figure 8-right shows the wear map obtained from optical CMM measurements (laser scanning using the multisensor CMM described in Section 2.3) of the same ceramic femoral head as in Figures 4 and 8-left. Initially the femoral head was measured using only the laser sensor, but this scan produced inaccurate results: the expanded uncertainty of volume measurement obtained in this case was above 0.5% of the sample volume, which is considered not accurate enough for measuring prosthesis wear volumes [7]. This was due mainly to large probing errors in correspondence with the most tilted portions of the measured surface respect to the laser beam (corresponding to the unworn areas used for datum alignment). The accuracy can be improved by multisensor measurements, using tactile probing for the alignment to the unworn areas, while the laser sensor is used only for probing the top calotte representing the worn area (see the coloured area in Figure 8-right, excluding the grey zone), as demonstrated in [7]. Figure 9 shows the wear map obtained from tactile CMM measurements of the UHMWPE non-congruous uni-condylar knee prosthetic component whose CT measurements were already presented in Section 3.1 (see Figure 7). The expanded uncertainty of volume measurement obtained in this case was 0.5% of the sample volume, which is more than the double of the uncertainty obtained by CT measurement (see the previous section).

iCT Conference 2014 – www.3dct.at 215

Figure 9: Deviation map showing wear distribution for a polyethylene non-congruous uni-condylar knee prosthetic

component (same sample as in Figure 7) measured by tactile CMM.

The considerable increment of uncertainty in CMM measurement that occurs passing from ceramic to polyethylene components (from 0.02% in the case of the ceramic head to 0.5% in the case of the UHMWPE knee component) was found due mainly to the elastic deformations introduced by the fixture system needed for clamping the samples on the CMM table. Figures 10 and 11 show the entity of such elastic deformations. These deviation maps were obtained by measuring new (unworn) polymeric components fixed on the CMM, and comparing the measured surface to the nominal surface representing the unloaded geometry.

Figure 10: Deviation map showing elastic deformation

due to fixation method on a new polyethylene non-congruous uni-condylar knee prosthetic component

measured by tactile CMM.

Figure 11: Deviation map showing elastic deformation due to fixation method on a new polyethylene

congruous uni-condylar knee prosthetic component measured by tactile CMM.

4 Discussion and conclusions This work demonstrated that CT measurements are capable of quantifying the wear and deformation of prosthetic components with volume measurement uncertainty (k=2) below 0.2% of the sample volume, for both ceramic and polymeric components. Comparisons with CMM measurements showed that tactile CMM probing may be more accurate than CT for the measurement of ceramic components, with expanded uncertainties that can be reduced down to 0.02% of the sample volume. However, tactile CMM suffers of limitations when measuring polymeric components, due to the probing and clamping forces that may influence the measured geometry. In the case of polyethylene prosthetic components, CMM measurements were affected by large errors introduced by elastic deformations caused by the fixture system, while CT measurements allowed definitely more accurate wear quantification (with relative uncertainty below 0.2% in the case of CT, against 0.5% in the case of CMM, for non-congruous UKP measurement). Furthermore, another disadvantage of CMM with respect to CT is that long scanning paths are needed in order to ensure a sufficient density of scanned points; this is not only time-consuming, but may cause wear (and possibly failure) of the probe tip, especially when measuring worn ceramic components.

216

In addition, optical CMM measurements were also tested. Like CT measurements, and unlike tactile CMM measurements, they do not need probing and clamping forces. However, other major limitations were found. In particular, large probing errors may occur especially in correspondence of tilted portions of the measured surface. Additional limitations are arising when measuring translucent or reflective parts, as can be the case for polymeric or metal components. CT was found more accurate than CMM laser scanning for measuring wear of the prosthetic components studied in this work. Finally, CT was tested also for measuring retrieved prosthetic components. Results were found similar to those obtained from non-retrieved parts, except from the influence of residuals, such as adhesive or bone, which may reduce the accuracy on datum alignment.

References

[1] G.E. Petsatodis, P.P. Papadopoulos, K.A. Papavasiliou, I.G. Hatzokos, F.G. Agathangelidis, A.G. Christodoulou, 2010. Primary cementless total hip arthroplasty with an alumina ceramic-on-ceramic bearing: results after a minimum of twenty years of follow-up. J Bone Joint Surg Am, 92(3):639-44.

[2] F. Prudhommeaux, M. Hamadouche, J. Nevelos, C. Doyle, A. Meunier, L. Sedel, 2000. Wear of alumina-on-alumina total hip arthroplasties at a mean 11-year followup. Clin Orthop Relat Res, (379):113-22.

[3] ISO 14242-1:2012. Implants for surgery – Wear of total hip-joint prostheses – Part 2: Loading and displacement parameters for wear-testing machines and corresponding environmental conditions for test. ISO, Geneve, 2012.

[4] L. Blunt, P. Bills, X. Jiang, C. Hardaker, G. Chakrabarty, 2009. The role of tribology and metrology in the latest development of bio-materials. Wear, 266:424-431.

[5] M. Spinelli, S. Carmignato, S. Affatato, M.Viceconti, 2009. CMM-based procedure for polyethylene non-congruous unicompartmental knee prosthesis wear assessment. Wear, 267: 753-756.

[6] S. Carmignato, M. Spinelli, S. Affatato, E. Savio, 2011. Uncertainty evaluation of volumetric wear assessment from coordinate measurements of ceramic hip joint prostheses. Wear, 270(9-10):584-590.

[7] S. Carmignato, E. Savio, 2011. Traceable volume measurements using coordinate measuring systems. CIRP Annals, vol. 60/1; p. 519-522,

[8] ISO 15530-3:2011. Geometrical product specifications (GPS) – Coordinate measuring machines (CMM): Technique for determining the uncertainty of measurement – Part 3: Use of calibrated workpieces or measurement standards. ISO, Geneve, 2011.

[9] W. Dewulf, K. Kiekens, Y. Tan, F. Welkenhuyzen, J.-P. Kruth, 2013. Uncertainty determination and quantification for dimensional measurements with industrial computed tomography. CIRP Annals, 62/1:535–538.

[10] J.-P. Kruth; M. Bartscher; S. Carmignato; R. Schmitt; L. De Chiffre; A. Weckenmann; 2011. Computed Tomography for Dimensional Metrology. Keynote paper. CIRP Annals, 60/2: 821-842.