Polyethylene wear represents a significant risk factor for the long-term success of knee arthroplasty [1]. This work aimed to develop and in vivo validate an automated algorithm for accurate and precise AI based wear measurement in knee arthroplasty using clinical AP radiographs for scientifically meaningful multi-centre studies. Twenty postoperative radiographs (knee joint AP in standing position) after knee arthroplasty were analysed using the novel algorithm. A convolutional neural network-based segmentation is used to localize the implant components on the X-Ray, and a 2D-3D registration of the CAD implant models precisely calculates the three-dimensional position and orientation of the implants in the joint at the time of acquisition. From this, the minimal distance between the involved implant components is determined, and its postoperative change over time enables the determination of wear in the radiographs. The measured minimum inlay height of 335 unloaded inlays excluding the weight-induced deformation, served as ground truth for validation and was compared to the algorithmically calculated component distances from 20 radiographs. With an average weight of 94 kg in the studied TKA patient cohort, it was determined that an average inlay height of 6.160 mm is expected in the patient. Based on the radiographs, the algorithm calculated a minimum component distance of 6.158 mm (SD = 81 µm), which deviated by 2 µm in comparison to the expected inlay height. An automated method was presented that allows accurate and precise determination of the inlay height and subsequently the wear in knee arthroplasty based on a clinical radiograph and the CAD models. Precision and accuracy are comparable to the current gold standard RSA [2], but without relying on special radiographic setups. The developed method can therefore be used to objectively investigate novel implant materials with meaningful clinical cohorts, thus improving the quality of patient care

Objectives

Numerous complications following total knee replacement (TKR) relate to the patellofemoral (PF) joint, including pain and patellar maltracking, yet the options for in vivo imaging of the PF joint are limited, especially after TKR. We propose a novel sequential biplane radiological method that permits accurate tracking of the PF and tibiofemoral (TF) joints throughout the range of movement under weightbearing, and test it in knees pre- and post-arthroplasty.

Anterior Cruciate Ligament (ACL) rupture is one of the commonest injuries in sports medicine. However, the rates of the reported graft re-rupture range from 2–10%, leading to around 3000 to 10000 revision ACL reconstructions in United States per annum. Inaccurate tunnel positions are considered to be one of the commonest reasons leading to failure and subsequent revision surgery. Additionally, there remains no consensus of the optimal position for ACL reconstructions. The positions of the bone tunnels in patients receiving ACL reconstruction are traditionally assessed using X-rays. It is well known that conventional X-ray is not a precise tool in assessing tunnel positions. Thus, there is a recent trend in using three-dimensional (3D) CT. However, routine CT carries a major disadvantage in terms of significant radiation hazard. In addition, it is both inconvenient and expensive to use CT as a regular assessment tools during the follow-up. The goal of the present work is to develop a novel 2D-3D registration method using single X-ray image and a surface model. By performing such registration for two post-operative X-rays, we can further calculate the 3D tunnel positions after ACL reconstructions. Our framework consists of five parts: (1) a surface model of the knee, (2) a 2D-3D registration algorithm, (3) a 3D tunnel position calculation, (4) a graphic user interface (GUI), and (5) a semi-transparency rendering. Among them, the crucial part is our 2D-3D registration method that estimates the relative position of the knee model in the imaging coordinate system. Once registered, the 3D position of an ACL tunnel in the knee model is calculated from the imaging geometry. The only interaction required is to mark the ACL tunnels on the X-rays through the GUI. We propose two 2D-3D registration methods. One is a contour-based method that uses pure geometric information. Most methods in this category accomplish the registration by extracting contours in X-rays, establishing their correspondences on the 3D model, and calculating the registration parameters. Unlike these methods, which need point-to-point correspondences, our method optimises the registration parameters in a statistical inference framework without giving or establishing point-to-point correspondences. Due to the use of the statistical inference, our method is robust to the spurs and broken contours that automatically extracted by the contour detector. The second method takes into account both the geometric shape of the object and the intensity property (intensity changes) of the image, where the intensity changes can be detected via image gradients. The use of gradient is based on the interpretation that two images are considered similar, if intensity changes occur at the same locations. The angles between the image gradients and the projected surface normals were used as a distance measure. The summation of the measures for all projected model points gives us the gradient term, which we multiply the contour-based measurement. Multiplication is preferred over addition because addition of the terms would require both terms to be normalised. To evaluate the feasibility of our methods, a simulation study was conducted using Digitally Reconstructed Radiographs (DRR) of a sawbone underwent a single-bundle ACL reconstruction performed by an experienced orthopedic surgeon. The real position of the bone tunnel entry point was obtained using the CT images, which were acquired using a custom-made well-calibrated cone-beam CT. The knee model was built by downsampling and smoothing the high-resolution CT reconstructions. It is important in our experiments to make the model different from the original reconstruction since this simulates the condition in which patient's CT is unavailable. Two DRRs generated from approximately anteroposterior and lateral viewpoints were used. For each DRR, 50 trials of 2D-3D registration were carried out for the femoral part using 50 different initialisations, which were randomly selected from the values independently and uniformly distributed within ±10 degrees and ±10 mm of the ground-truth. Compared with the ground-truth established using the CT images, our single image contour-based method achieved accurate estimations in rotations and in-plane translations, which were (−0.67±1.38, −0.98±0.84, −0.42±0.71) degrees and (0.11±0.26, −0.06±1.20) mm for the anteroposterior image, and (−0.78±0.76, −0.37±0.87, 0.70±0.88) degrees and (−0.14±0.22, 0.31±0.71) mm for the lateral one, respectively. The same experiments were also performed using the second method. However, it did not produce desirable results in our experiments. The tunnel entry point was then calculated using the averaged registration result of our contour-based method. The entry point of the tunnel was obtained with high accuracy of 1.25 mm distance error from the real position of the entry point. For the 2D-3D registration, the estimated off-plane translations showed relatively low accuracy. It is well known that the depth can be difficult to be accurately estimated using one single image. As the result showed, the accuracy in rotations and in-plane translations is more important for ACL tunnel position estimation in our framework. As for the image gradient, it is too sensitive to the small perturbation caused by image noises. A more robust way of integrating the gradient information into our contour-based method is required. We propose a novel approach for estimating the 3D position of bone tunnels in ACL reconstruction using two post-operative X-rays. It was tested in a sawbone study using DRRs. The most significant advantage of our approach is to potentially eliminate the necessity of acquiring a patient's CT. The success in developing and validating the proposed workflow will allow convenient and precise assessment of tunnel positions in ACL reconstruction with minimal risk of radiation hazard

Summary. Bi-plane Image matching method is very useful technique to evaluate the loaded 3D motion of each cervical level. Introduction. Cervical orthoses are commonly used to regulate the motion of cervical spines for conservative treatment of injuries and for post-operative immobilization. Previous studies have reported the efficacy of orthoses for 2D flex-extension or 3D motions of the entire cervical spine. However, the ability of cervical orthoses to reduce motion might be different at each intervertebral level and for different types of motion (flexion-extension, rotation, lateral bending). The effectiveness of immobilizing orthoses at each cervical intervertebral level for 3D motions has not been reported. The purpose of this study is to evaluate the effectiveness of the Philadelphia collar to each level of cervical spines with 3D motion analysis under loading condition. Patients & Methods. Patient Sample: Four asymptomatic volunteer subjects were recruited and provided informed consent. Approval of the experimental design by the institutional review board was obtained. These 4 individuals were without any history of cervical diseases or procedures. The presence of any symptoms, spinal disorders and anatomical abnormalities in fluoroscopic images or CT was a criterion of exclusion from this study. Outcome Measures: To evaluate the efficacy of the Philadelphia collar, ANOVA was used to compare the range of motion with and without collar at the C3/4, C4/5, C5/6 and C6/7 intervertebral levels for each motion. The level of statistical significance was set at p<0.05. When a statistical difference was detected, post hoc Tukey tests were performed. Methods. Three-dimensional models of the C3-C7 vertebrae were developed from CT scans of each subject using commercial software. Two fluoroscopy systems were positioned to acquire orthogonal images of the cervical spine. The subject was seated within the view of the dual fluoroscopic imaging system. Pairs of images were taken in each of 7 positions: neutral posture, maximum flexion and extension, maximum left and right lateral bending, and maximum left and right rotation. The images and 3D vertebral models were imported into biplane 2D-3D registration software, where the vertebral models were projected onto the pair of digitised images and the 3D bone pose was adjusted to match its radiographic projection in each image. Relative motions between each vertebral body were calculated from body-fixed coordinate systems using a flexion-lateral bending-axial rotation Cardan angle sequence. Results. Flexion range was significantly reduced with the collar at each cervical level. Extension range was significantly reduced at the C3/4 level. Rotation and lateral bending were reduced for C3/4, C4/5, C5/6 levels with the collar. Discussion/Conclusion. The Philadelphia Collar significantly reduces cervical motion at C3/4, C4/5 and C5/6 levels in almost all motions (except for extension). At the C6/7 level, this type of collar has limited effectiveness reducing cervical motion. We used 3D radiographic measurements to quantify the effectiveness of the Philadelphia collar for reducing cervical motion. Bi-plane 2D-3D registration method is useful technique to evaluate 3D motion of cervical spines

Introduction. Cervical orthoses are commonly used to regulate the motion of cervical spines for conservative treatment of injuries and for post-operative immobilization. Previous studies have reported the efficacy of orthoses for 2D flex-extension or 3D motions of the entire cervical spine. However, the ability of cervical orthoses to reduce motion might be different at each intervertebral level and for different types of motion (flexion-extension, rotation, lateral bending). The effectiveness of immobilizing orthoses at each cervical intervertebral level for 3D motions has not been reported. The purpose of this study is to evaluate the effectiveness of the Philadelphia collar to each level of cervical spines with 3D motion analysis under loading condition. Patients & Methods. Patient Sample Four asymptomatic volunteer subjects were recruited and provided informed consent. Approval of the experimental design by the institutional review board was obtained. These 4 individuals were without any history of cervical diseases or procedures. The presence of any symptoms, spinal disorders and anatomical abnormalities in fluoroscopic images or CT was a criterion of exclusion from this study. Outcome Measures To evaluate the efficacy of the Philadelphia collar, ANOVA was used to compare the range of motion with and without collar at the C3/4, C4/5, C5/6 and C6/7 intervertebral levels for each motion. The level of statistical significance was set at p < 0.05. When a statistical difference was detected, post hoc Tukey tests were performed. Methods. Three-dimensional models of the C3–C7 vertebrae were developed from CT scans of each subject using commercial software (see Figure 1). Two fluoroscopy systems were positioned to acquire orthogonal images of the cervical spine. The subject was seated within the view of the dual fluoroscopic imaging system (see Figure 2). Pairs of images were taken in each of 7 positions: neutral posture, maximum flexion and extension, maximum left and right lateral bending, and maximum left and right rotation. The images and 3D vertebral models were imported into biplane 2D-3D registration software, where the vertebral models were projected onto the pair of digitized images and the 3D bone pose was adjusted to match its radiographic projection in each image (see Figure 3). Relative motions between each vertebral body were calculated from body-fixed coordinate systems using a flexion-lateral bending-axial rotation Cardan angle sequence. Results. Flexion range was significantly reduced with the collar at each cervical level. Extension range was significantly reduced at the C3/4 level. Rotation and lateral bending were reduced for C3/4, C4/5, C5/6 levels with the collar. Discussion/Conclusion. The Philadelphia Collar significantly reduces cervical motion at C3/4, C4/5 and C5/6 levels in almost all motions (except for extension). At the C6/7 level, this type of collar has limited effectiveness reducing cervical motion. We used 3D radiographic measurements to quantify the effectiveness of the Philadelphia collar for reducing cervical motion. Bi-plane 2D-3D registration method is useful technique to evaluate 3D motion of cervical spines

The standard approach for kinematic analysis of knee joints has been roentgen stereophotogrammetry (RSA). This approach requires implanting tantalum beads during surgery so pre- and post-surgery comparisons have not been conducted. CT- fluoroscopy registration is a non-invasive alternative but has had accuracy and speed limitations. Our new algorithm addresses these limitations. Our approach to the problem of registering CT data to single-plane fluoroscopy was to generate a digitally reconstructed radiograph (DRR) from the CT data and then filter this to produce an edge-enhanced image, which was then registered with an edge-enhanced version of the fluoroscopy frame. The algorithm includes a new multi-modal similarity measure and a novel technique for the calculation of the required gradients. Three lower limb specimens were implanted with 1 mm tantalum beads to act as fiducial markers. Fluoroscopy data was captured for a knee flexion and femur and tibia CT data was registered to the fluoroscopy images. A previous version of our algorithm (developed in 2008) showed good accuracy for in-plane translations and rotations of the knee bones. However, this algorithm did not have the ability to accurately determine out-of-plane translations. This lack of accuracy for out-of-plane translations has also been the major limitation of other single-plane 2D-3D registration algorithms. Fregly et. al. and Dennis et. al. reported standard deviations for this measurement of 5.6 and 3.03 mm respectively. The latest version of our algorithm achieves error standard deviations for out-of-plane translations of 0.65 mm. The algorithm includes a new similarity measure, which calculates the sum of the conditional variances (SCV) of the joint probability distributions of the images to be registered. This new similarity measure determines the true 3D position of the bones for a wider range of initial disparities and is also faster than the cross-cumulative residual entropy (CCRE) measure used in the 2008 version. For a set of initial 3D positions ranging from ± 5 pixels and ± 5 degrees the proposed approach successfully determined the correct 3D position for 96% of cases–whilst the approach using CCRE was successful for only 49% of cases. The algorithm also required 60% less iterations than the previous CCRE approach. The new registration algorithm developed for the project provides a level of accuracy that is superior to other similar techniques. This new level of accuracy opens the way for a non-invasive mechanism for sophisticated kinematic analysis of knee joints. This will enable prospective, longitudinal and controlled studies of reconstruction surgery

Purpose. Measurements of patellar kinematics are essential to investigate the link between anterior knee pain following knee arthroplasty and patellar maltracking. A major challenge in studying the patellofemoral (PF) joint postoperatively is that the patellar component is only partially visible in the sagittal and close-to-sagittal radiographs. The narrow angular distance between these radiographs makes the application of conventional bi-planar fluoroscopy impossible. In this study a methodology has been introduced and validated for accurate estimation of the 3D kinematics of the PF joint post-arthroplasty using a novel multi-planar fluoroscopy approach. Method. An optoelectronic camera (Optotrak Certus) was used to track the motion of an ISO-C fluoroscopy C-arm (Siemens Siremobil) using two sets of markers attached to the X-ray source and detector housings. The C-arm was used in the Digital Radiography (DR) mode, which resembles an ordinary X-ray fluoroscopy image. A previously-developed technique (Cho et al., 2005; Daly et al., 2008) was adapted to find the geometric parameters of the imaging system. Thirty-eight DRs of the calibration phantom were obtained for the 190 of rotation of the C-arm at 5 rotational increments while data from motion markers were recorded continuously at a frequency of 100 Hz. A total knee replacement prosthesis was implanted on an artificial bone model of the knee, and the implant components and bones were rigidly fixed in place using a urethane rigid foam. For the purpose of validation, positions of the implant components were determined using a coordinate measuring machine (CMM). Sagittal and obliquely sagittal radiographs of the model were taken where the patellar component was most visible. For each DR the geometric parameters of the system were interpolated based on the location of the motion markers. The exact location of the projection was then determined in 3D space. JointTrack Bi-plane software (Dr. Scott Banks, University of Florida, Gainesville) was used to conduct 2D-3D registration between the radiographs and the reverse-engineered models of the implant components. Results of the registration were directly compared to the ground-truth obtained from the CMM to calculate the accuracies. Results. The accuracies for the PF were found to be 0.48 mm and 1.32 for position and orientation of the components. For the tibiofemoral joint these values were found to be 0.89 mm and 1.43, respectively. Conclusion. The multi-planar method can be used to assess the sequential kinematics of the patellofemoral and tibiofemoral joints including the mediolateral translation and tilt of the patellar component, which are obscured in standard 2D sagittal measurements and are not possible using the traditional bi-planar setup. A limitation is that it can only be used for static imaging of the joint. It has the advantage of a relatively low radiation dose. This methodology can be used to investigate the relationship between maltracking of the patella and anterior knee pain as well as other postoperative complications

Introduction. The goal of this work is to develop a system for three-dimensional tracking of the acetabular fragment during periacetabular osteotomy (PAO) using x-ray images. For PAO, the proposed x-ray image-based navigation provides geometrical and biomechanical assessment of the acetabular fragment, which is unavailable in the conventional procedure, without disrupting surgical workflow or requiring tracking devices. Methods. The proposed system combines preoperative planning with intraoperative tracking and near real-time automated assessment of the fragment geometry (radiographic angles) and biomechanics (contact pressure distribution over the acetabular surface). During PAO, eight fiducial beads are attached to the patient after incision and prior to performing osteotomy. Four of the beads attach to the iliac wing above the expected superior osteotomy (these are termed confidence points), and four attach on the expected fragment (denoted fragment points). At least two x-ray images are obtained before and after osteotomy. In each set of images, image processing routines segment the fiducials and triangulate the 2D fiducial projections in 3D space. A paired-point registration between the confidence points triangulated from the two x-ray image sets aligns the imaging frames. We measured the transformation between the fragment points with respect to the confidence points to quantify the motion of the acetabular fragment. Applying an image-based 2D-3D registration to the measured acetabular transformation localises the reoriented acetabular fragment with respect to an anatomical coordinate system. We present the surgeon with visualisation and automatic estimations of radiographic angles and biomechanics of the reoriented acetabular fragment. We conducted an experiment to evaluate feasibility and accuracy of the proposed system using a high density pelvic sawbone. Stainless steel beads were glued to the sawbone as fiducials. X-ray images were selected from cone-beam CT (CBCT) scans with an encoded motorised C-arm. Fiducial segmentation from reconstructed volumes of the CBCT scans provided a ground truth for the experiment. Results. We used four images spaced at 45° to perform the 2D/3D registration. The measured fragment transformation errors in translation and rotation about a fixed axis when compared to the CBCT-computed transformation were 0.37°, 0.34mm for the x-ray image based approach (with 3 images spaced at 20°) and 1.49°, 4.39mm for the optical tracker. Conclusion. We developed and evaluated x-ray image-based navigation to track the acetabular fragment in 3D Cartesian space during PAO. Capturing the fragment transformation allows automated algorithms to assess the biomechanics and geometry of the realigned acetabulum that are unavailable in 2D. The error between the measured positions of the beads and the triangulated positions is attributed to three main sources: 1) fiducial segmentation error; 2) geometric calibration error; and 3) localisation of fiducials in volumetric reconstructions of the CBCT scans. These small reported errors suggest the procedure is a viable approach for conducting x-ray image-based navigation of PAO

Introduction and Aims: A method has been designed to accurately measure post-operative alignment of hip (acetabular) and knee (femoral and tibial) prosthetic components relative to the pre-operative plan. Conventional methods involve 2D measurements; this new method uses 2D-3D registration to align both the prosthesis and the pre-operative CT volume to the post-operative x-ray. Method: The method uses an automatic approach to align a CAD model of the prosthesis to the post-operative x-ray. A rendering of the prosthesis is produced and overlaid onto the post-operative x-ray image. The prosthesis can be rotated and translated in 3D to match the outline of the rendering shown on the post-operative x-ray. An initial manual procedure is used to align the rendering of the bone surface from pre-operative CT to the bony anatomy on the post-operative x-ray. This manual registration position is then used as a starting position for an automated intensity-based registration algorithm. Results: The automated intensity-based registration algorithm allowed 3D verification of the prosthesis position. A number of digitally reconstructed radiographs (DRRs) were produced by casting rays through the pre-operative CT volume. The DRRs were then compared with the post-operative X-ray image using a similarity measure. This similarity measurement was optimised using gradient decent-type search strategy to alter the rotation and translation parameters. If the Hounsfield numbers of the voxels, which the casting rays passed through, were integrated along the ray and projected onto an imaging plane, a radiograph-like image was produced. To concentrate the area of registration and thus quicken registration algorithm, the user also manually defined a region of registration interest. Hence, DRRs were only produced within the region of interest. Due to the large size of the pelvis and tube-like nature of the femur and tibia, a total of 10 starting positions were used for this algorithm. These starting positions were found by adding random Gaussian noise to the parameters found using the manual process. The registration position was defined as the final position that produced the best similarity measurement value. Conclusion: Validation has demonstrated this method’s accuracy in calculating the post-operative position of acetabular and knee prostheses with respect to the pre-operative plan. The results are repeatable, robust and enable pre- and post-operative 3D implant position comparison. The inaccuracies observed with conventional methods due to incorrect alignment on x-ray are reduced

The long-term result of a total hip arthroplasty (THA) strongly depends on the correct component positioning of the acetabular cup and stem. To measure cup orientation out of a postoperative anteroposterior (AP) pelvic radiograph is highly inaccurate due to the wide variation of individual pelvic tilt and rotation. The goal of this study was to develop and validate a 2D-3D matching software (HipMatch) that allows matching a postoperative AP pelvic radiograph with a preoperative CT to accurately measure cup orientation corrected for individual pelvic orientation. The software is based on a spline-based multi-resolution 2D-3D image registration algorithm and a Markov random field theory based on similarity measurement. Based on a cone projection (imitating the path of the x-rays), the software is able to match the three-dimensional CT-based data set with the contours of the projected pelvis on the AP pelvic radiograph. This gives the possibility to correct the measured cup orientation (inclination and anteversion) by measuring it according to an anatomical defined coordinate system (anterior pelvic plane). The validation of the software consisted of accuracy, reproducibility and observer reliability measurements using cadaver and clinical data. For the cadaver validation 10 human pelves (20 hips) were used. From each pelvis 2 CT scans, one with and one without an inserted cup were acquired. The CT scan with the cup was used as the ground truth. With the cup inserted 4 AP pelvic radiographs with the pelvis in an unknown arbitrary position during acquisition were performed resulting in 80 measurements for accuracy. These measurements were performed by 2 observers at 2 different occasions resulting in a total of 320 measurements for reproducibility and observer reliability. The intraclass correlation coefficient (ICC) was used for quantification of reproducibility and observer reliability and the Bland-Altman analysis was used to detect systemic errors. The clinical validation included 33 patients with a pre- and a postoperative CT and 49 patients with only a postoperative CT in addition to the postoperative radiographs. In the cases with only a postoperative CT, for the 2D-3D matching the postoperative CT after manual excision of the cup from the CT slice sticks was used. In all cases the postoperative CT was used as the ground truth. For each patient all the available postoperative radiographs were used resulting in 236 measurements of accuracy. In the cadaver validation the cup orientation ranged from 34° – 57° for the inclination and from 1° – 24° for the anteversion measured on the CT. The accuracy showed a mean difference for the inclination of 0.9° ± 1.6° (−3.2° – 4.0°) and of 1.2 ± 2.4° (−5.3° – 5.6°) for the anteversion. The ICC for the reproducibility ranged from 0.96 to 0.99 and for the interobserver reliability from 0.95 to 0.98. No relevant systematic error was detected. In the clinical validation the cup orientation measured on the postoperative CT ranged for the inclination from 22° – 57° and for the anteversion from 7° – 35°. In the clinical setup the accuracy showed a mean difference for inclination of 1.8° ± 1.6° (−4.0° – 5.3°) and of −1.1° ± 2.9° (−5.9° – 5.7°) for the anteversion. The 2D-3D matching technique showed a good accuracy and a very good reproducibility and observer reliability. This technique allows to measure the exact cup orientation out of an AP pelvic radiograph with the help of a preoperative CT and to correct the parameters for the individual pelvic position. Therefore this software is a powerful tool to measure accuracy of CT-based computer-assisted cup placement in a large clinical series