Simpleware Case Study: Statistical Shape Modeling of the Large Acetabular Defect in Hip Revision Surgery

Introduction

Total hip arthroplasty (THA) is a common orthopedic procedure for restoring hip joint function when severe acetabular defects are present. However, managing severe acetabular bone loss is challenging for revision THA using traditional methods, often leading to high revision rates due to implant failure.

Custom-made acetabular implants help overcome these obstacles by fitting the implant to the residual host bone, thus helping more accurately position acetabular components for improved performance and a smaller risk of revision surgery. Methods for planning these surgeries include mirroring the healthy contralateral hemipelvis to measure an acceptable hip center of rotation (CoR), but are limited by human anatomy asymmetry, pathologies, or metal work.

While pelvic Statistical Shape Models (SSMs) provide a solution for reconstructing pelvic bone defects, these have mainly been limited to computational models without relation to patient data, or based on very small patient cohorts. This case study involved researchers going further with SSMs to test their value in the reconstruction of important bony landmarks that are absent in hip joints of patients with large acetabular defects. By using this method, it is possible to compare the CoR of the SSM with patients who had previously undergone hip revision surgery for large acetabular defects planned without the SSM technique.

Study Design and Outcome Measures

The dataset used for this project was a retrospective cohort study involving 38 patients with Paprosky type IIIb defects, including major destruction of the acetabular rim and teardrop. The SSM was trained using 50 healthy pelvises, and then used to virtually reconstruct the native pelvic morphology for all 38 cases, of which 18 had healthy contralateral sides.

The SSM could then be compared with the preoperative CT-based plan for all patients whose surgeries were planned without the SSM technique, and outcome measurements found for the difference between CoR between the SSM and diseased hip, the SSM and the preoperative plan, and the SSM and the contralateral healthy hip. This analysis shows how improvements could be made to future surgical planning.

Data Preparation in Simpleware Software

One hundred pelvic CT scans of patients (fifty male and fifty female) without bony abnormalities were imported to Simpleware software and automatically segmented using AI-based Machine Learning tools to obtain a 3D reconstruction for each hemipelvis. The pre-operative CT scans of the Paprosky patient cohort were also segmented in Simpleware software using a combination of manual and automatic segmentation tools.

The 3D virtual plans of the 38 cases were used to test the statistical model using the contralateral hemipelvises where an implant was not present, with two of the 38 not undergoing revision surgery and only having the virtual plan. The validation set was formed of 10 healthy hemipelvises from the training set, and was used to validate the virtual reconstruction process.

Creating the Statistical Shape Model (SSM)

50 pelvises from the training set were initially aligned to ensure a fixed pose of the dataset. The mean shape of the hemipelvis was then registered to each hemipelvis using a point mapping technique, with the mean shape as a reference object for providing the locations of the points to be mapped. Principal Component Analysis (PCA) was applied on the dataset to investigate the correlations between the mean shape and the model set.

The study assumed that the healthy parts of the diseased hips from the test set could predict the shape of the anatomy, whereby the acetabulum was removed and the remaining parts used to build the SSM. The healthy parts were aligned to the best fit model generated from the mean shape, as well as the dense data points of the original geometries.

Image Analysis and Validation

The CoR of the SSM was calculated using a sphere matching technique, with the same method used for computing the CoR of the diseased hip, the plan, and the healthy contralateral side. The difference in CoR between each diseased hip and its respective SSM was calculated to quantify the severity of the defect. Finally, the difference in CoR between the SSM and the plan, and the SSM and the healthy side, was calculated. The coordinate system was changed with respect to the anterior pelvic plane using anatomical definitions, reference points employed to create a plane, and centroid distances calculated.

Statistical analysis was performed using GraphPad Prism to investigate statistically significant differences between the groups. Data was determined to be non-normally distribution, and the Mann-Whitney U test for non-parametric independent values subsequently carried out, with the significance level set at p > 0.05. The SSM built using the training set was used to virtually reconstruct the hemipelvises of 10 healthy patients, and their CoR measured via the sphere matching technique. Differences in CoR with respect to the corresponding SSM were calculated to test the model's performance as a planning tool.

Results and Discussion

Image analysis and validation results reviewed the discrepancy in CoR between the diseased hip and the SSM, the plan and the SSM, and the healthy contralateral side and the SSM; this final metric was used to validate the model. The largest difference was found between the diseased hips and corresponding SSMs (p < 0.0001), with no statistical differences found in terms of CoR between the plan and the SSM, and contralateral vs. SSM difference in CoR. The validation showed that the largest displacement in CoR was found in the coronal plane.

Analysis found that the CoR of the pathological hips were within 31 mm of the SSM reconstructed model, on average; the CoR of the surgical plan was located within 9 mm of the SSM model, excluding seven cases where the surgeon chose to retain a high centre of rotation and not restore native CoR. The CoR of the healthy contralateral side was within 8 mm of the SSM, with no statistical differences found between the plan vs. SSM and contralateral vs. SSM difference in CoR for patients with planned restoration of native CoR.

Results show promise for using the SSM as a reference tool for guiding surgical planning of hip revision surgery where the contralateral side is diseased or has an implant. Future studies may address a larger patient cohort to better assess whether the SSM can cope with different types of severe acetabular defects or deformities.

Conclusions

Hip reconstruction in patients with large acetabular defects is challenging due to their lack of anatomical landmarks, the result of the deformed anatomy and the contralateral hip being often replaced. In addition, failed implants can create metal artefacts that obscure bony readings from CT scans, making 3D reconstruction challenging. The SSM workflow in this case study overcomes these limitations by successfully reconstructing the absent bony landmarks of diseased anatomies, regardless of the severity of the defects; this is important for estimating the original position of the CoR prior to the development of the bony defect, which can then be used as a starting point for engineers to design customized implants that are less likely to fail or produce complications for patients with large acetabular defects.

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