Simpleware Case Study: Statistical Shape Modeling the In Vivo Location of Acetabular Wear in Retrieved Hip Implant

Mean acetabular surface generated through a PCA of the population (CC BY 4.0)


Edge-wear in acetabular cups is known to correlate to greater volumes of material loss, but the location of this wear pattern in vivo is less understood. A workflow using CT imaging, retrieval analysis and statistical shape modelling (SSM) in Simpleware software has been developed to identify the most common locations of wear in vivo. 

A shape variance study was conducted from twenty retrieved metal-on-metal acetabular surfaces that were revised after a mean time period of 90 months. The study also looked at the impact of wear volume, positioning, time, size, and gender on the in vivo location of wear, providing insights that can help to better understand hip implant function, informing future designs and the refinement of a safe zone for implant positioning.


  • 20 retrieved metal-on-metal acetabular surfaces were scanned to capture wear patterns
  • Statistical shape modelling (SSM) in Simpleware software used to identify the most common acetabular wear pattern features generated in vivo 
  • SSM enables improved understanding of hip implant function to inform future design decisions and refine safe zones for implant positioning

"The statistical shape modelling tools developed by Synopsys within Simpleware software have opened up novel and useful insights into analyzing retrieved orthopedic implants, enabling us to better understand their function in-vivo. The capabilities used here show great promise for helping us advance future implant design and ultimately improving patient outcomes."

Johann Henckel MD, Researcher
Royal National Orthopaedic Hospital


Mechanical wear at the bearing surface of hip replacements can affect clinical performance, leading to impaired function and the release of harmful debris. With high incidents of metal-on-metal (MOM) hips failing, retrieval studies have been carried out to investigate the extent of wear on these surfaces which have identified a relationship between wear on the acetabular cup edge and high volumes of wear debris. While MOM implants are now scarcely used in hip implants, they still provide valuable data for analyzing the mechanics of hip replacements. In addition, although it is known that acetabular edge wear occurs in vivo, there is less understanding of its orientation within the acetabular cavity. 

Statistical shape models (SSM) offer a valuable method for describing the shape and position of a population of related geometries, particularly when analyzing anatomical features in such a way that the mean shape and shape variance within a cohort can be visualized. In this study, the goal was to identify the most common in vivo wear patterns present in the acetabular component of hip replacements by combining CT imaging and retrieval analysis techniques to create an SSM in Simpleware software.

Study design and workflow of acetabular wear in retrieved hip implant (CC BY 4.0)

Study design and workflow, accompanied by the backside surface of a Birmingham acetabular component and its asymmetrical stabilising fins (circled in red) (Image by Bergiers et al. / CC BY 4.0).

Acetabular components of twenty retrieved MOM Birmingham hip replacements (BHRs) were selected for the study based on inclusion criteria that required pre-revision 3D CT images showing the pelvic bone and implant prior to removal. The implants had been revised by two of the study authors, following a mean time in vivo of 90 months, with surgery carried out either due to adverse reactions to metal debris, unexplained pain, or aseptic loosening. 

The in vivo position of each BHR was calculated using a bespoke software solution (Robin’s 3D), which used the anterior pelvic plane (APP) as a standardized coordinate system and reported values of anatomical inclination and anteversion. The articulating surface geometry of each retrieved cup was then captured as point clouds using a Carl Zeiss coordinate measuring machine (CMM), while the volume of material loss from each acetabular surface was calculated using a previously validated, automated software solution.

Segmentation and Registration in Simpleware Software

The pre-revision CT images of the in vivo implants were imported as DICOM files to Simpleware software and segmented using semi-automated tools to generate implant and pelvic bone models. Minimal postprocessing was used to reduce the presence of metal artefacts, while retaining the geometric accuracy. A sphere was then best-fit to the femoral head and subtracted through a Boolean operation to isolate the acetabular cup from the implant model, before a plane was best-fit to the cup rim to remove the relatively inferior portion of the model, including the remanence of the femoral peg. Open surface representations of the acetabular cups, generated from the CMM data, were imported as STL files and registered to the isolated cup models, with a semi-automated function to align their stabilized fins. The registered acetabular cup surfaces and bone models were subsequently mirrored and appropriately scaled.

Acetabular surface registered to the 3D model of its acetabular cup (CC BY 4.0)

(1) An acetabular surface (purple) registered to the 3D model of its acetabular cup (grey), segmented form its 3D CT images. (2) All 20 acetabular surfaces registered using the CAPP axis, maintaining their in vivo orientation. (3) The CAPP axis (red) used to divide the acetabular surface into four quadrants (Image by Bergiers et al. / CC BY 4.0).

Registration of the BHR surfaces using the anterior pelvic plane was enabled by aligning all twenty acetabular surfaces through a standardized coordinate system, defined using a plane parallel to the anterior pelvic plane (APP) that intersected the center of the cup surface. The plane was called the Cup-APP (CAPP) as being representative of the vertical standing position due to its relationship to the APP. A new coordinate system was then created to establish final and measurement axes from which the in vivo location of the primary wear scar could be determined, and the acetabular surface was split into four quadrants (anterosuperior, anteroinferior, posterosuperior, and posteroinferior).

CAPP axis defined using a plane parallel to the APP (CC BY 4.0)

The CAPP axis was defined using a plane parallel to the APP (left image), which intersected the centre of the sphere formed by the acetabular surface (right image) (Image by Bergiers et al. / CC BY 4.0).

Statistical Shape Modelling

To carry out SSM, the aligned acetabular surfaces were clipped at the cup-rim transition, and capped to form closed surfaces, before each surface was discretized to form dense point sets for mapping to a reference unworn geometry. Simpleware software was used to perform a principal component analysis (PCA) and to generate a shape model that could be morphed from the mean acetabular surface geometry through the identified models of variation (principal components).

Surface deviation maps were generated to compare the mean acetabular surface model with the reference unworn sphere and presented the location of wear as a map based on deviation values above manufacturing tolerances. The same approach was used to investigate the dominant modes of variance within the population of acetabular surfaces, and a leave-one-out study was performed to evaluate the contribution of each acetabular surface to the overall statistical shape model.

Box and Whiskers plot presenting the location of the primary wear scar centre (CC BY 4.0)

Box and Whiskers plot presenting the location of the primary wear scar centre, measured after each iteration of the leave-one-out study. The median is presented along with the range (min-max) and interquartile range. The black dot represents the single outlier (Image by Bergiers et al. / CC BY 4.0).

The study also considered the impact of factors such as the surgeon, implant type, and patient, as well as the influence of volumetric material loss, gender, and time to revision. The twenty acetabular surfaces were therefore divided into groups to investigate their differences. The same method was used to calculate the wear volume of each acetabular surface from the surface deviation maps generated by Simpleware software, and the results were compared with measurements obtained by a previously validated software. The mean error between the two volumes was found to be acceptable.


The mean acetabular surface was generated through a PCA of the whole implant population, presented above as a deviation map of their comparison with the as-manufactured reference surface. In addition, the leave-one-out study indicated that a single acetabular surface significantly influenced the center of the mean wear scar and was consequently excluded from further analyses.

Mean acetabular surface generated through a PCA of the entire population (CC BY 4.0)

The mean acetabular surface generated through a PCA of the entire population (n = 20). This is presented as deviation map of their comparison with the as-manufactured reference surface, where the dark blue regions are considered unworn (mm) (Image by Bergiers et al. / CC BY 4.0).

PCA generates modes, which are eigenvectors that describe the shape variance within the population and are ranked based on the degree of variance in their direction. The first and most dominant mode of variance generated from PCA reflects the linear depth at the primary wear scar, the second mode describes the relationship between the scar location and its coverage towards the center of the surface, and the third mode indicates the degree of pinching along the horizontal axis.

Change in geometry from −3SD to +3SD presented as deviation maps (CC BY 4.0)

The change in geometry from −3SD to +3SD of the first 3 principal components (modes) calculated from this population of acetabular surfaces, presented as deviation maps of their comparison with the as-manufactured reference surface (mm) (Image by Bergiers et al. / CC BY 4.0).

Simpleware software was also used to replicate the recorded material loss measurements, compared to the center and radius of the implant’s unworn spheres calculated by the previously validated software, within a mean error of -0.60mm3 (SD = 6.98). Please see the full paper for more details on the results of this study.


This project is notable for being the first study to use statistical shape modelling for the analysis of retrieved orthopedic implants. The mean in vivo location of acetabular wear was successfully identified, with SSM able to visualize and interpret shape variance within a population of retrieved resurfacing hips. In addition, the method produced results consistent with findings from the literature, but with an improved insight into the mechanics of hip replacements as a way of informing future implant designs and the study of pathological anatomies. The SSM method can also potentially be used to refine the safe zone for implant positioning and help inform surgeons of the range of component positions that lead to optimal wear performance, with the potential for further effectiveness through larger population studies.


Except where otherwise noted, content on this page is licensed under a Creative Commons Attribution 4.0 International license.

Any Questions?

Do you have any questions about this case study or how to use Simpleware software for your own workflows?