Introduction

In clinical routine surgeons depend largely on 2D x-ray radiographs and their experience to plan and evaluate surgical interventions around the knee joint. Numerous studies have shown that pure 2D x-ray radiography based measurements are not accurate due to the error in determining accurate radiography magnification and the projection characteristics of 2D radiographs. Using 2D x-ray radiographs to plan 3D knee joint surgery may lead to component misalignment in Total Knee Arthroplasty (TKA) or to over- or under-correction of the mechanical axis in Lower Extremity Osteotomy (LEO).

Recently we developed a personalized X-ray reconstruction-based planning and post-operative treatment evaluation system called “iLeg” for TKA or LEO. Based on a patented X-ray image calibration cage and a unique 2D–3D reconstruction technique, iLeg can generate accurate patient-specific 3D models of a complete lower extremity from two standing X-rays for true 3D planning and evaluation of surgical interventions at the knee joint. The goal of this study is to validate the accuracy of this newly developed system using digitally reconstructed radiographs (DRRs) generated from CT data of cadavers.

Recently we developed a personalised X-ray reconstruction-based planning and post-operative treatment evaluation system called iLeg for total knee arthroplasty or lower extremity osteotomy. Based on a patented X-ray image calibration cage and a unique 2D-3D reconstruction technique, iLeg can generate accurate patient-specific 3D models of a complete lower extremity from two standing X-rays for true 3D planning and evaluation of surgical interventions at the knee joint. The goal of this study is to validate the accuracy of this newly developed system using digitally reconstructed radiographs (DRRs) generated from CT data of 12 cadavers (24 legs). Our experimental results demonstrated an overall reconstruction accuracy of 1.3±0.2mm.

Tracked B-mode ultrasound (US) potentially provides a non-invasive and radiation-free alternative to percutaneous pointer digitization for intra-operative determination of the anterior pelvis plane (APP). However, most of the published approaches demand a direct access to the corresponding landmarks, which can only be presumed for surgical approaches with the patient in supine position. In order to avoid any change of the clinical routine for total hip arthroplasties (THAs), we propose a new method to determine the pelvic orientation, which could be performed in lateral position.

Our proposed method is based on the acquisition of ultrasound images of the ipsilateral hemi-pelvis, namely the posterior superior iliac spines (PSISs) and iliac crest region. The US images are tracked by a navigation system and further processed to extract three-dimensional point clouds. As only one side of the pelvis is accessible, we estimate the symmetry plane (midsagittal plane) of the pelvis based on additionally digitized bilateral anterior superior iliac spine (ASIS) landmarks. This symmetry plane is further used to mirror the ipsilateral US-derived points to the contralateral side of the pelvis and to register and instantiate a pelvic SSM constructed from 30 CT-scans.

The proposed registration method was evaluated using two plastic pelvis models and two cadaveric pelvises together with special custom-made silicone phantoms to simulate the missing soft-tissue. In each trial, the required data were collected with the pelvis rigidly fixed in lateral decubitus position together with ground truth APP landmarks. A registration error of 3.48° ± 1.10° was found for the anteversion angle, while the inclination angle could be reconstructed with a mean error of 1.26° ± 1.62°.

The performed in-vitro experiments showed reasonably good results, taking the sparsity of the input point clouds into consideration.

Bone fixation plates are routinely used in corrective and reconstructive interventions. Design of such implants must take into consideration not only good surface fit, but also reduced intra-operative bending and twisting of the implant itself. This process increases mechanical stresses within the implant and affects its durability and the functional outcome of the surgery. Wound exposure and anaesthesia times are also reduced. Current population-based designs consider the average shape of a target bone as a template to pre-shape the implant. Other studies try to enhance the average design by optimising surface metrics in a statistical shape space. This could ensure a low mean distance between the implant and any bone in the population, but does not reduce neither the maximum possible distances nor directly the mechanical forces needed to fit the implant to the specific patient. We propose a population-based study that considers the bending and torsion forces as metrics to be minimised for the design of enhanced fixation plates. Our aim is to minimise the necessary intra-operative deformations of the plates.

In our approach, we first propose to represent a fixation plate by dividing it into discrete sections lengthwise and fitting a plane to each section. The number of sections depends on the size of the implant and anatomical location. It should be small enough to capture the anatomical curvatures, but large enough not to be affected by local noise in the surface.

Surface patches corresponding to common locations for plate fixations are extracted from 200 segmented computed tomography (CT) images. In this work, distal lateral femoral patches are considered. A statistical shape model of the patches is then computed and a large population of 2,197 instances is generated, evenly covering the natural statistical variation within the initial population. These instances are considered as both bone surfaces and potential new designs of the contact surface of the fixation plate.

The key formulation of our solution is to examine the effect of deforming each section of the implant on the rest of the sections and compute the amount of bending and torsion needed to shape one patch to another.

Each instance of the population is fitted to all others and the maximum bending and torsion angles are recorded. A similar process was applied for the mean of the population. The goal is to pick from the population the shape that simultaneously minimises the bending and torsion angles.

The maximum required bending was reduced from 25.3® to 19.3® (24.72% reduction), whereas the torsion component was reduced from 12.4® to 6.2® (50% reduction).

The method proposed in this abstract enhances the current state-of-the-art in orthopaedic implant design by considering the mechanical deformations applied to the implant during the surgery. The obtained results are promising and indicate a noticeable improvement over the standard pre-contouring to the population mean. We plan to further validate the method and as a future outlook, we intend to test the approach in real surgical scenarios.

The integration of statistical shape models (SSMs) for generating a patient-specific model from sparse data is widely spread. The SSM needs to be initially registered to the coordinate-system in which the data is acquired and then be instantiated based on the point data using some regressing techniques such as principal component analysis (PCR). Besides PCR, partial least squares regression (PLSR) could also be used to predict a patient-specific model. PLSR combines properties of PCR and multiple linear regression and could be used for shape prediction based on morphological parameters.

Both methods were compared on the basis of two SSMs, each of them constructed from 30 surface models of the proximal femur and the pelvis, respectively. Thirty leave-one-out trials were performed, in which one surface was consecutively left out and further used as ground truth surface model. Landmark data were randomly derived from the surface models and used together with the remaining 29 surface models to predict the left-out surface model based on PCR and PLSR, respectively. The prediction accuracy was analysed by comparing the ground truth model with the corresponding predicted model and expressed in terms of mean surface distance error.

According to their obtained minimum error, PCR (1.62 mm) and PLSR (1. 63 mm) gave similar results for a set of 50 randomly chosen landmarks. However PLSR seems to be more susceptible to a wrong selection of number of latent vectors, as it has a more variation in the error.

Although both regression methods gave similar results, decision needs to be done, how to select the optimal number of regressors, which is a delicate task. In order to predict a surface model based on morphological parameters using PLSR, the choice of the parameters and their optimal number needs to be carefully selected.

The goal of this study was to validate accuracy and reproducibility of a new 2D/3D reconstruction-based program called “HipRecon” for determining cup orientation after THA. “HipRecon” uses a statistical shape model based 2D/3D deformable registration technique that can reconstruct a patient-specific 3D model from a single standard AP pelvic X-ray radiograph. Required inputs include a digital radiograph, the pixel size, and the film-to-source distance. No specific calibration of the X-ray, or a CAD (computer-assisted design) model of the implant, or a CT-scan of the patient is required. Cup orientation is then calculated with respect to the anterior pelvic plane that is derived from the reconstructed 3D-model.

The validation study was conducted on datasets of 29 patients (31 hips). Among them, there were 15 males and 14 females. Each dataset has one post-operative X-ray radiograph and one post-operative CT-scan. The post-operative CT scan for each patient was used to establish the ground truth for the cup orientation. Radiographs with deep centering (7 radiographs), or of pelvises with fractures (2 radiographs), or with both (1 radiograph), or of non-hemispherely shaped cup (1 radiograph) were assessed separately from the radiographs without above mentioned phenomena (18 radiographs) to estimate a potential influence on the 2D/3D reconstruction accuracy. To make the description easier, we denote those radiographs with above mentioned phenomena as non-normal cases and those without as normal cases. The cup anteversions and inclinations that were calculated by “HipRecon” were compared to the associated ground truth. To validate the reproducibility and the reliability, one observer conducted twice measurements for each dataset using “HipRecon”.

The mean accuracy for the normal cases was 0.4° ± 1.8° (−2.6° to 3.3°) for inclination and 0.6° ± 1.5° (−2.0° to 3.9°) for anteversion, and the mean accuracy for the non-normal cases was 2.3° ± 2.4° (−2.1° to 6.3°) for inclination and 0.1° ± 2.8° (−4.6° to 5.1°) for anteversion. Comparing the measurement from the normal radiographs to those from the non-normal radiographs using the Mann-Whitney U-test, we found a significant difference in measuring cup inclination (p = 0.01) but not in measuring cup anteversion (p = 0.3). Bland-Altman analysis of those measurements from the normal cases indicated that no systematical error was detected for “HipRecon,” as the mean of the measurement pairs were spread evenly and randomly for both inclination and anteversion. “HipRecon” showed a very good reproducibility for both parameters with an intraclass correlation coefficient (ICC) for inclination of 0.98 (95% Confidence Limits (CL): 0.96–0.99) and for anteversion of 0.96 (95% CL: 0.91–0.98).

Accurate assessment of the acetabular cup orientation is important for evaluation of outcome after THA, but the inability to measure acetabular cup orientation accurately limits one's ability to determine optimal cup orientations, to assess new treatment methods of improving acetabular cup orientation in surgery, and to correlate the acetabular cup orientation to osteolysis, wear, and instability. In this study, we showed that “HipRecon” was an accurate, consistent, and reproducible technique to measure cup orientation from post-operative X-ray radiographs. Furthermore, our experimental results indicated that the best results were achieved with the radiographs of non-fractured pelvises that included the anterior superior iliac spines and the cranial part of the non-fractured pelvis. Thus, it is recommended that these landmarks should be included in the radiograph whenever the 2D/3D reconstruction-based method will be used

The existing image-free Total Hip Arthroplasty (THA) navigation systems conventionally utilise the patient-specific Anterior Pelvic Plane (APP) as the reference to calculate orientations of the implanted cup, e.g. anteversion and inclination angles. The definition of APP relies on the intra-operative digitisation of three anatomical landmarks, the bilateral Anterior Superior Iliac Spine (ASIS) and the pubicum.

Due to the presence of the thick soft tissue around the patient's pubic region, however, the landmark on pubic area is hard to be digitised accurately. A novel reference plane called Intra-operative Reference Plane (IRP) was proposed by G. Zheng et al to address this issue. To determine the IRP, bilateral ASIS and the cup center of the operating side instead of the pubicum are digitised intra-operatively. It avoids the error-prone digitisation of pubicum, and the angle between the patient-specific APP and the suggested IRP can be computed pre-operatively by a single X-ray radiograph-based 2D/3D reconstruction approach developed by G. Zheng et al. Based on this angle, the orientation of the APP can be intra-operatively estimated from that of the IRP such that all measurements with respect to IRP can be transformed to measurements with respect to APP.

In order to implement and validate this new reference plane for image-free navigation of acetabular cup placement, we developed an IRP-based image-free THA navigation system. All cup placement instruments were mounted with passive markers whose positions could be traced by a NDI Polaris® infrared camera (Northern Digital Inc, Ontario, Canada). The cup center was obtained by first pivoting a tracked impactor with appropriate size of the mounted trial cup and then calculating the pivoting center through a least-squares fitting. The bilateral ASIS landmarks were acquired through the percutaneous pointer-based digitisation.

We tested this new IRP-based image-free THA navigation system in our laboratory by conducting twelve studies on two dry cadaver pelvises and two plastic pelvises. The ground truth for each study was established using the conventional APP-based method, i.e., in addition to those landmarks required by our IRP-based method, we also digitised the pubicum on respective pelvic bones and calculated cup orientations on the basis of the digitised APP.

The mean and standard deviation of differences between the proposed IRP-based anteversion measurement and the ground truth are 1.0 degree and 0.7 degree, while the maximal and minimal differences are 2.1 degree and 0.3 degree respectively.

The mean and standard deviation of differences between the proposed IRP-based inclination measurement and the ground truth are respective 0.2 degree and 0.2 degree. Moreover, the maximum of differences is 0.5 degree and the minimum is 0.0 degree.

Our laboratory experimental results demonstrate that the new IRP-based image-free navigation system is accurate enough for acetabular cup placement. In comparison to existing image-free navigation systems that use APP as the reference plane, the newly developed system employs IRP as the reference plane, which has the advantage to eliminate the digitisation of landmarks around the pubic region. The successful validation with the laboratorial study has led us to the next step of clinical trials. We expect to report preliminary clinical cases in the near future.