The analysis of defects in custom 3D‐printed acetabular cups: A comparative study of commercially available implants from six manufacturers

Three‐dimensional (3D) printing is used to manufacture custom acetabular cups to treat patients with massive acetabular defects. There is a risk of defects occurring in these, often in the form of structural voids. Our aim was to investigate the presence of voids in commercially available cups. We examined 12, final‐production titanium custom acetabular cups, that had been 3D‐printed by six manufacturers. We measured their mass, then performed micro‐computed tomography (micro‐CT) imaging to determine their volume and density. The micro‐CT data were examined for the presence of voids. In cups that had voids, we computed (1) the number of voids, (2) their volume and the cup volume fraction, (3) their sphericity, (4) size, and (5) their location. The cups had median mass, volume, and density of 208.5 g, 46,471 mm3, and 4.42 g/cm3, respectively. Five cups were found to contain a median (range) of 90 (58–101) structural voids. The median void volume and cup volume fractions of cups with voids were 5.17 (1.05–17.33) mm3 and 99.983 (99.972–99.998)%, respectively. The median void sphericity and size were 0.47 (0.19–0.65) and 0.64 (0.27–8.82) mm, respectively. Voids were predominantly located adjacent to screw holes, within flanges, and at the transition between design features; these were between 0.17 and 4.66 mm from the cup surfaces. This is the first study to examine defects within final‐production 3D‐printed custom cups, providing data for regulators, surgeons, and manufacturers about the variability in final print quality. The size, shape, and location of these voids are such that there may be an increased risk of crack initiation from them.

The use of three-dimensional (3D) printing, also known as additive manufacturing, is becoming more common in the manufacture of titanium spine, hip, and knee implants. The clinical rationale is the potential for enhanced bony fixation within highly porous implant surfaces, and the ability to print implants with complex shapes to treat patients with challenging bony anatomies. 1 In hip arthroplasty, 3D printing is increasingly being used to manufacture custom acetabular cups for patients with massive acetabular defects. 2,3 Whilst this technology is rapidly being adopted by the orthopedic industry, other industries, such as the aerospace sector, have been more cautious. This is due, in part, to the known risk of structural defects that may occur during the printing process. 4 These defects most often present as voids within the component but may also be seen as cracks or contamination of the starting metal powder. 5 The potential impact of defects is a risk of fracture of the printed part, depending on their size and location. 6 A previous study using micro-computed tomography (micro- CT) imaging has shown that some off-the-shelf 3D-printed acetabular cups contain structural voids. 7 There is no evidence to suggest that the structural integrity of these cups is compromised and indeed there are no known clinical reports of fractures of these components.
It has been shown however that the risk of voids occurring increases as the complexity of the shape being printed increases 8 ; we do not currently know the impact of this on custom acetabular cups that are becoming more widely used.
In this study we examined 12 final-production, custom acetabular cups that had been 3D-printed by six manufacturers. Our aim was to use micro-CT imaging to characterize any void defects that existed within these components.

| Materials
We examined 12 unused, final-production titanium alloy acetabular cups that had been 3D-printed with a custom design to treat patients with large acetabular defects, Figure 1. This implant group consisted of two designs printed by each of the six leading manufacturers of 3D-printed orthopedic implants. Each design was made up of the main acetabular body and at least one flange. All the cups examined had exceeded the 6-month window in which they could be used in the patients they had been designed for.
In this study, each of the six manufacturers was randomly assigned with an identifier between Cup_1 and Cup_6; the two cups from each manufacturer were randomly assigned a label of A or B.
F I G U R E 1 Macroscopic images of the 12 cups examined in this study, which were 3D-printed by six manufacturers. Two views of the front and back of each cup are presented.
From this point on, the implants will be referred to as Cup_1A, Cup_1B, Cup_2A, Cup_2B, and so forth.
The study design is summarized in Figure 2.

| Cup mass
We measured the mass of each cup using Mettler PC 4400 (Mettler Toledo) digital scales.

| Micro-CT imaging
We used a Nikon XTH 225 micro-CT scanner (Nikon Metrology) to perform high-resolution 3D imaging of each cup. Scanning was performed with a beam current and voltage of up to 150 μA and 225 kV, respectively, with the components as close as possible to the beam source whilst still capturing the entire component in the field of view. A 1 mm thick copper filter was positioned at the beam source to minimize the beam hardening effects that occur when scanning a metal sample. A total of 3177 frames were captured, with an exposure of 1000 ms, in increments of 0.11°. Figure 3 illustrates the analysis steps performed.

| Reconstruction of micro-CT data
CT Pro 3D software (Nikon Metrology) was used to reconstruct the two-dimensional (2D) projection images. A filtered back-projection algorithm was utilized, incorporating second-order polynomialcorrection numerical filtering to further minimize any beam hardening that may have occurred.

| Cup volume
The filtered and corrected micro-CT data were imported into the analysis software package Volume Graphics and the cups were F I G U R E 2 Summary of the analysis steps that were performed, in which (a) the final-production 3D-printed implant was obtained, (b) the implant was imaged using a micro-CT scanner and reconstructed in 3D, (c) a void detection algorithm was applied to the 3D reconstruction, (d) the voids were identified and confirmed as likely defects, and (e) the volume, sphericity, size, and location of each void were computed. micro-CT, micro-computed tomography.
segmented using an ISO-50 approach. An automated calculation of the volume of each cup was then performed, before the examination of potential voids within the components.

| Cup density
A calculation of the density of each cup was made as mass/volume, in which the measure of volume assumed no voids to be present.

| Void analysis
The Porosity/Inclusion Analysis module within the Volume Graphics software was then used to investigate the presence of any voids within the components, indicating structural defects. The VGDefX algorithm was applied during analysis, in which variations in grayscale values are assessed and compared for each voxel of the scan to determine if the voxel forms part of a void or of the material. An automatic deviation threshold mode was used, with a deviation factor of 0.7 was applied for the standard deviation of the titanium material peak of the gray value distribution. Low noise reduction was applied to the data.
An edge distance calculation was performed for each void that was identified, to determine the minimum distance between the void and the surface of the implant. Following this analysis, the following parameters were obtained for each cup.

| Number of voids
We recorded the total number of voids that were identified as being likely defects within each component and also calculated the number of voids/mm. 3

| Void volume
We recorded the volume (mm 3 ) of each individual void and the total volume of voids for each cup. We used this to determine the volume fraction, which is a measure of the percentage of material in each cup relative to its volume.

| Void sphericity
We recorded the sphericity of each void, calculated as a ratio between the surface area of a sphere with the same volume as the void and the surface area of the void itself.

| Void size
The size (mm) of each void was determined by measuring the diameter of the smallest sphere that could encompass the entire void.

| Void location
The transparency of the titanium material in each cup was reduced to 90% to better visualize the highlighted voids within the components; these were then inspected to identify their locations. The edge distance calculations were used to report the distance of each void from the cup surface.

| RESULTS
We found that Cups 1A, 1B, 3A, 3B, and 5B had evidence of structural defects, presenting as voids within the components. The remaining cups appeared to be free of such voids, Figure 4. Table 1 summarizes all the defect analysis data that were obtained in this study. Figures 5-9 illustrate the cavities that were identified in five out of the 12 cups.

| Cup mass
The median (range) mass of the cups was 208.5 (133-419) g.

| Cup volume
The median (range) volume of all cups in this study was 46,471 (30,264-94864) mm 3 .
F I G U R E 3 Summarizing the study design used

| Void volume
The cups with voids were found to have a median volume of 5.17

| Void sphericity
The median sphericity of all voids identified in this study was 0.47 (0.19-0.65), Figure 11.

| Void size
The median size of all voids in the cups was 0.64 (0.27-8.82) mm,

| DISCUSSION
This is the first study to investigate the presence of structural defects within final-production 3D-printed custom acetabular cups. We found that five out of the 12 cups examined had evidence of internal T A B L E 1 Summary of the defect analysis data obtained for each cup, detailing the presence, and characteristics of structural voids  The median sphericity of the voids identified was 0.47 which indicates that were more irregularly shaped rather than being more spherical (i.e., with a sphericity of closer to 1.0). It has been suggested that larger, irregular voids are more prone to stress risers than smaller spherical voids, which may have a negative impact on the mechanical performance of a component. 14  Nevertheless, our study did find that 7 out of the 12 cups examined had no evidence of structural voids that were detectable at the scanning resolution used (i.e., no voids greater than 0.09 mm were detected within these seven cups). The voids that were found in the current study appear to have primarily been formed due to a "lack All manufacturers utilize postprocessing methods to eliminate voids that may form during the printing process. The most commonly used method is known as hot isostatic pressing (HIP) in which a high-isostatic-pressure gas, usually argon, is applied to the printed part at temperatures high enough to maximize plastic deformation of the material, theoretically leading to internal voids collapsing. 16 Subsequent processes of creep and diffusion lead to voids fully closing and being entirely eliminated. Experimental studies have shown the HIP process to be highly effective at removing large voids within the central regions of printed parts however it has been reported that this method may be less effective at fully eliminating voids closer to the surface. 8 The findings from our study are consistent with these experimental studies, in which the voids identified were all located a median of 0.5 mm from the implant surfaces. We do not know however the precise postprocessing methods that were used by the different manufacturers and if the absence of voids in some designs was due to a difference in the postprocessing or indeed in the printing process itself.
We acknowledge the limitations of this study. Whilst our micro-CT parameters were optimized to maximize the resolution of the scans, we were not able to detect voids that were smaller than twice the voxel size of the scans (i.e., smaller than 2 × 0.045 mm = 0.09 mm). It may be the case that smaller cavities were present in these cups however, as discussed previously, it has been suggested that these sizes are unlikely to impact the mechanical integrity of 3D-printed components. 8 Interestingly, the smallest void that we detected was 0.28 mm in size, notably larger than the minimum detectable size. Nevertheless, future studies should seek to understand if smaller cavities are present in these types of implants and involve a greater number of implants to understand the scale of print variability between manufacturers.
Future research should also extend to independent mechanical testing of these components to better understand failure mechanisms arising from these voids.

| CONCLUSION
This is the first study to examine structural defects within finalproduction 3D-printed custom acetabular cups. We found evidence of voids in five cups that were between 0.27 and 8.82 mm in size and located between 0.17 and 4.66 mm from their surfaces. The size, shape, and location of these voids are such that there may be an increased risk of crack initiation from them. This study provides data for regulators, surgeons, and manufacturers about the variability in print quality between different designs in relation to eliminating printing defects.