CAOS 2020: Papers with Abstracts

Abstract. The specific navigation system used can be currently considered as a reliable instrument for measuring the kinematics of a TKA for HKA angle and medial and lateral gaps, but non for femoral translation and rotation.

Abstract. Achieving a balanced total knee throughout the entire range of motion leads to improved patient reported outcomes and satisfaction. Sensor-assisted technology allows the surgeon to quantitatively assess and address imbalance through either soft tissue releases or bone recuts. However, balancing through soft tissue releases leads to unpredictable gap increments and frequent early over-releases.

METHODS: During a consecutive and prospective series of 29 robotic total knee surgeries, intra-operative load sensors were used following the initial bone resections to quantitatively assess the knee’s state of balance through the range of motion with trial components in place. Load measurements were taken at 10 and 90 degrees of knee flexion. Based on previous literature, a balanced knee is defined as having a mediolateral load difference below 15 pounds (lbf) through the range of motion, with an absolute load magnitude per compartment above 5lbf and not exceeding 45lbf. The initial load numbers were recorded as well as the number and type of subsequent corrections needed to achieve quantitative balance.

RESULTS Of the 29 robotics cases, only 12 (41%) were well-balanced after the initial bone cuts (mechanical alignment by measured resection). Another two cases were too loose and required an increase in the polyethylene thickness size of two millimeters to achieve a well-balanced knee without further bone resection. In 14 cases, a bone recut was required to balance the knee. More specifically, four cases required a recut of the femur, ten cases required a recut of the tibia. Eventually, one case was left unbalanced in flexion with a mediolateral load differential of 20 lbf. It should be noted explicitly that no soft tissue releases were done for any of the 29 cases. At the end, all 29 knees were considered well balanced in extension and all but six (79%) at 90° of flexion. For these six cases with balance issue at 90° of flexion, absolute load magnitude in both compartments was below 45 lbf and above 5lbf, though the mediolateral load differential was between 15lbf and 30lbf.

DISCUSSION Based on a preliminary series, this work demonstrates the opportunity of combining multiple technologies to achieve a quantitatively balanced knee through the range of motion without any soft tissue release.

Abstract. The recent studies about iliosacral screws performed with navigation systems show promising results. The Surgivisio system is a new generation of intraoperative 3D imaging technique used in our institution since two years. The aim of this prospective study was to evaluate the accuracy of iliosacral screw placement and radiation exposure with the Surgivisio® system.
Between January 2018 and December 2019, every patient operated for percutaneous iliosacral screwing using the Surgivisio® system were included in this prospective single center study. Accuracy of screw placement was assessed with post-operative high- resolution CT-scan. Operative time, radiation exposure and complications were assessed.
A total of 32 patients were included with 49 iliosacral screws. Using the modified Gras classification, 2% (1/49) were rated as misplaced and 2% (1/49) were repositioned. The mean operative time was 26 min for the whole procedure. The mean dose area product was 7.98 Gy.cm2. Two complications were recorded (neurological pain treated by removal of the misplaced screw, an asymptomatic cement leakage with one augmented iliosacral screw).
The Surgivisio® system is an efficient navigation tool for iliosacral screwing in minimal invasive surgery. It improves the accuracy of screw placement with an acceptable radiation exposure and operative time.

Abstract. The first patient-specific biomechanical model for pre-operative planning in acetabular surgery was developed in our institution and validated in previous retrospective studies. The aim of this prospective clinical study was to confirm the previous promising.
Between January 2019 and June 2019, every patients operated by the first author for acetabular fracture were included in this prospective study. A biomechanical model was implemented in acustom software made from combination of several open-sources software allowing a biomechanical simulation. The surgery was then performed according to the simulation. Surgery duration, blood loss, radiological results and per- operative complications were recorded.
Ten patients were included. Mean simulation time was 22 min ± 4 [range, 18-31]. The mean operative time was 113 min ± 33 [range, 60-180] and mean blood loss was 505 mL ± 189 [range, 100-750]. On Matta’s criteria, anatomic reduction was achieved in 9 of the 10 patients (90%; 0.8 mm ± 1 [range, 0-3]). No per-operative complications were recorded.
This study confirms the promising results of pre-operative planning in acetabular surgery based on a patient-specific biomechanical model.The model needs larger-scale prospective validation, but offers a new tool suitable for teaching purposes and for assessment of surgical strategies in acetabular fracture.

Abstract. In the last years also in orthopedic surgery, there was an increasing interest in the development of surgical simulators using methods of additive manufacturing combined or not with augmented reality systems (hybrid simulators). Aim of this work was to evaluate the use of a new patient’s specific tibial plateau fractures simulator for surgical training of young resident surgeons in fracture fixation with an external fixator. The simulator is a realistic knee phantom including a patient-specific replica of a fractured tibia and fibula, obtained by CT segmentation and rapid prototyping techniques. Each training session started with the presentation, and planning, of the surgical case that it was followed by the external fixation session on the simulator. At the end of each session, all participants were asked to fill out a questionnaire, concerning the phantom realism and appropriateness as a teaching modality. The results of the Likert Questionnaire indicating that there is an overall significant agreement with the phantom realism and its appropriateness as a teaching modality.
The solid model of the patient’s anatomy can faithfully reproduce the surgical complexity of the patient and it allows to generate surgical simulators with an increasing difficulty to perform structured training paths: from the "simple" case to the "complex" case. The use of simulators based on 3D models has proved to be a very useful tool both for didactic and surgical training purposes, allowing surgeons to perform a real procedure simulation outside the surgical room.

Abstract. Femoral component loosening due to poor fixation is a common failure mode in cementless total hip arthroplasty (THA). We sought to develop a technique to predicted femoral stem fixation using pre-operative CT.
Forty-nine patients requiring THA received pre-operative CTs as part of Corin OPSTM 3D dynamic planning. All patients received a taper wedged blade stem (TriFitTM TS, Corin UK) implanted through a posterior approach. Post-operatively, patients received an immediate CT and AP x-ray and a 1-year follow-up x-ray. The distance between the most superior point of the greater trochanter and the shoulder of the stem was measured and scaled using the known cup diameter on both the immediate and 1- year follow-up x-rays. The difference was recorded as stem subsidence. Subsidence greater than 4mm was deemed clinically relevant. The precise placement of the stem determined by registering the known 3D implant geometry to the post-op CT was virtually implanted back into the pre-operative OPSTM planning software. A colour map of the bone density at the stem/bone interface using the Hounsfield Units (HU) of each CT pixel was generated. Blue represents low density bone transitioning through to green and then red.
Mean stem subsidence was 1.8mm (0 to 11.1mm). Five patients had clinically relevant subsidence. Significant areas of blue and green around the proximal portion of the stem was observed in high subsidence stems when compared to the stems with minimal subsidence.
Using the HU of the CT scan as an indicator for bone density, we were able to predict poor implant fixation and subsequent subsidence in a taper wedge stem.

Abstract. Aims To evaluate: (1) objective quantification of ligament balancing in total knee arthroplasty, (2) types and effectiveness of additional procedures to compartment pressure, and (3) change of pressure values in both compartment throughout the range of motion in total knee arthroplasty.
Methods Eighty-four patients underwent total knee arthroplasty (TKA) using VERASENSE Knee System. TKA was performed by two techniques. Compartment pressure was recorded through the range of motion (ROM) initially, after each additional procedure, and after final implantation. Balanced knees were defined as when the compartment pressure difference was less than 15 pounds.
Results Thirty patients (35.7%) showed “balanced” knee on initial pressure measurement. Modified gap balancing TKAs showed significantly higher proportion of “balanced” knee than measured-resection TKAs (P = 0.004). Both medial and lateral compartment pressure were generally decreased on both TKA methods. Linear correlation showed statistically significant through ROM on both compartment. Total 66 additional ligament balancing procedures were performed.
Conclusion Using the objective orthosensor, we were able to obtain 94% of well-balanced total knee arthroplasty finally. Furthermore, acquired objective data can lead to proper ligament balancing for both experienced and young surgeons and consequently reduce the complications associate with soft tissue imbalance in the future.

Abstract. This study employed an advanced method (CUSUM) to analyze the learning curve regarding surgical efficiency (time) using two CAOS applications, which were designed to address user needs with different levels of comprehensiveness in term of offered guidance and instrumentation requirements. Two group of surgeons, each used either CAOS applications were included in the study. The first 50 CAOS TKA cases from each surgeon were analyzed to identify the learning curve. The duration of learning, as well as the impact of learning based on surgical time, were assessed with regard to the specific CAOS application and surgeon’s previous CAOS experience level. The data demonstrated differences in term of pattern of adoption during learning process between the two CAOS applications. However, the learning process was not sensitive to surgeon’s experience level.

Abstract. Principal component analysis on 376 TKA knees presented the distinctive patterns and variabilities in the coronal knee angular laxity throughout the range of motion, measured post-implantation during computer-assisted total knee arthroplasty. The variability in the laxity curves were dominated by the first mode of variation (varus/valgus offset of the laxity envelope) and the second mode of variation (varus/valgus crossing pattern). Further analysis revealed that surgeon-specific impact was associated with the first mode of variation for the laxity curve, and both first and second mode of variation for the size of the laxity envelope. The results shed a light on the characteristics and variabilities of post-implantation soft-tissue laxity under surgical reality and may be used to further understand the clinical implications of intraoperative soft-tissue management.

Abstract. This study sought to evaluate the efficiency, usage, and accuracy of a novel technology that augments mechanical instrumentation with intraoperative CAOS guidance. Technical reports on 411 primary TKA cases performed using the technology were reviewed. The results demonstrated high surgical efficiency (time) and resection accuracy (alignment and resection depth). Furthermore, it was observed that one fifth of the time, the surgeons placed the cutting block that deviated more than >2°/mm from the ideal position in the coronal plane. Substantial adjustments were found to be required (on average ~10 °/mm per case) to correct the initial placement of the cutting block. The CAOS augmentation minimized the error in cutting block placement and assisted in achieving high accuracy in bony resections. The findings revealed the prevalence of clinical error with manual conventional bony preparation that can be addressed with efficiency and accuracy by adding CAOS augmentation to the mechanical instrumentation.

Abstract. The accuracy of implantation using navigation and robotic hip replacement (THR) has been proven and accepted. Time taken for the use of technology, has been one of the major factors quoted by many surgeons for not using it. Aim was to analyse the time taken during different steps of the procedure and identify which part needs further improvement. Secondary aim was to compare the time with computer navigation technology. 20 total joint replacements were carried out with MAKO robotic system using extended software version, which includes both acetabular and femoral guidance. The times of different surgical steps were noted. After the incision was made, and joint exposed the trackers for the pelvis and femur were then inserted, tightened and verification pins inserted. Femoral registration matching was started at 15 minutes (SD5.4 Range7-22). Femoral Neck Cut level was made at 22.5 minutes (SD4.5 Range15-29). It took mean 7.5 minutes to match the femoral anatomy to preop CT scan anatomy. Pelvic registration matching was done at 42.4 minutes (SD5.9 Range33-50). It took average 8.4 minutes to match pelvic anatomy. Bone reaming was with robotic arm at 50.8minutes (SD5.5 Range40-57). The cup was impacted at 56.2 minutes (SD5.2 Range45-60). Cementing stem added 14 minutes. The navigation hip replacement surgery took mean 60 minutes in uncemented and 72minutes for cemented stems. Robotic technique took 10 minutes longer. There is room for improvement in two main modifiable steps. Total registration matching time which took 16 minutes and using a uncemented stem. Should time constraint supersede accuracy achievable with technology?

Abstract. The accuracy of implantation using computer navigation and robotic total knee replacement (TKR) has been proven. Time taken during surgery has been a factor for surgeons for not using the technology. Aim of this study was to analyse time taken in different steps and identify which part needs improvement. Robotic time was compared to computer navigation. Methods: 15TKR were performed with MAKO robot. Software for the ligament balancing was used. All had CT scan preoperatively. Time of different surgical steps was recorded. Time for computer navigation was recorded too. After joint exposure, trackers and verification pins for tibia and femur were inserted. Femoral registration matching started at 10.8 minutes (SD3.3 Range7-20). It took 3.2 minutes to match femoral anatomy to CT scan. Tibial registration done at 14.1minutes (SD3 Range10- 23). Once matching was accepted to required accuracy, tibial cut was made at 22.2 minutes (SD4.4 Range 15-30). Next the soft tissues were assessed with tensioner. It took 6.3 minutes (SD 5.6). Final femoral preparation done at 35.7 minutes (SD 5.6 Range25-45). Trial performed at 52 minutes (SD7.3 Range42-63). Implants were cemented at 63.4 minutes (SD8 Range50-72). Wound closed at 77.6 minutes (SD9.5 Range65-97). The computer navigated TKR surgery took 70 minutes on an average. Compared to navigation, robotic technique took approximately 7 minutes longer, not significantly different. This could be due to learning curve of the surgical and theatre team. Improvement is required in different steps. The familiarity of staff will increase the efficiency. Registration matching took 11.4 minutes. Femoral preparation took 17 minutes. These steps could be streamlined.

Abstract. Accuracy of implantation using computer assisted navigation and robotic total knee replacement arthroplasty (TKR) and total hip replacement (THR) has been proven. Templating the pre-operative radiographs has become standard. This gives an insight into the pre op planning and the sizes most likely to be used. This helps to reduce the inventory for storage of implants and cost. While the templating of radiographs has been helpful, implant sizes prediction remains less than desirable. Aim of present study was to look at the predictability of implant sizes in CT planning for robotic surgery. 30 MAKO robotic joint replacements were performed (15TKR/15THR) with pre op CT scans for implant size. For TKR, the sizes used were mean 5 in femur and tibia (SD1 and range 3-7). In tibia, size used was same as predicted. In the femur in two cases the size was reduced by one to balance the gaps. Insert thickness was increased by one size in 4 cases. For THR, the acetabular cup, femur, head diameter and offset were predicted 100%. The neck length had to be changed in some cases by up to two sizes. for balance/stability. In conclusion the CT scan pre-operative planning for MAKO robotic knee joint replacement can predict 100% times the size of tibia, within one size of femur and insert for TKR. These figures are better than published predictability of templates of plain radiographs where implant size was predicted 42% for femoral and 37% acetabular components while 87% of the femoral components and 78% of the acetabular cups were accurate within one size up and down.

Abstract. Total knee arthroplasty (TKA) is a common procedure that has become the standard of treatment for severe cases of knee osteoarthritis. Biomechanics and quality of movement similar to healthy were found to improve patient-reported outcomes.
In this study, an evaluated musculoskeletal model predicted ligament, contact and muscle forces together with secondary tibiofemoral kinematics. An artificial neural network applied to the musculoskeletal model searched for the optimal implant position in a given range that will minimize the root-mean-square-error (RMSE) between post- TKA and native experimental tibiofemoral kinematics during a squat.
We found that, using a cruciate-retaining implant, native kinematics could be accurately reproduced (average RMSE 1.47 mm (± 0.89 mm) for translations and 2.89° (± 2.83°) for rotations between native and optimal TKA alignment). The required implant positions changes maximally 2.96 mm and 2.40o. This suggests that when using pre- operative planning, off-the-shelf CR implants allow for reproducing native knee kinematics post-operatively.

Abstract. Today, Implemented Medical Devices (IMDs) are key elements of the healthcare system being more and more complex with connectivity facilities. Participating to data collecting, they contribute to improve patient follow-up as well as medical practices. However, with such a connectivity security threats are significant and can be the cause of decision errors, patient privacy concerns and so on. This is why, international and national regulations make mandatory to consider security when developing new IMDs. In the context of resource constrained connected Knee Prosthesis (KP), we present: 1) a solution to securely process data emitted by KP on untrusted Human Machine Interface (HMI) and implementation times; 2) an authentication protocol for installing trust between the different entities involved in the process.

Abstract. During a typical fluoroscopic guided surgery, it is common to acquire multiple x-ray images to correctly position the C-arm. This can be a long process resulting in an in- crease in operation time and ionizing radiation exposure. Our purpose in this study is to implement a machine learning system for predicting the position of the C-arm based on the intraoperative radiographs. The prediction is achieved by training a Deep Learning Network based on Digitally Reconstructed Radiographs. We first showed a high prediction accuracy (4.5 mm and 1.1o) when patient-specific training was implemented. Additionally, we demonstrated a similar range of accuracy by applying transfer-learning on the last lay- ers of the network while reducing the processing time by 83%. In conclusion, in this study, we propose a C-arm position prediction system based on machine learning that can po- tentially reduce the number of intraoperatively acquired X-rays in a common orthopaedic surgical procedure.

Abstract. This study examined the advantages of CAOS TKA in short-term functional outcomes compared to conventional instrumented TKA. Data were collected from a prospective, multi-center study where 334 were treated with CAOS TKA, and 461 were treated with conventional TKA. Postoperatively at 2-3 year, the CAOS cases had significantly better range of motion (ROM), KSS function, and sub-component measures of KSS function than the conventional cases. Further analysis revealed that geographic region was associated with ROM but had no significant impact on KSS function. CAOS TKA was associated with higher postoperative KSS function and its sub-component measures. The short-term results reported support the use of CAOS technology with the goals of better function.

Abstract. This paper presents a projector-based augmented reality (AR) system for Computer- Assisted Orthopaedic Surgery (CAOS). After calibration, our AR system allows for projection of not only the virtual model directly on the surface of the target organ to create an augmented reality but also important clinical information such as distance and angular deviations from a surgical plan, which are important for various computer-assisted surgical procedures such as trajectory drilling and fracture reduction. The feasibility and accuracy of the system is experimentally validated on a 3D printed phantom model with pyramid shape, a dry goat bone and an in vitro pig leg. An average projection distance error of 1.03±0.58mm and an average drill alignment error of 1.17±0.43°were found. The results demonstrate the efficacy of the proposed AR system.

Abstract. In the context of automatic landmarks localization with statistical shape models for the design of customized TKA prosthesis, the first step consists of registering a model, represented by the mean mesh of some healthy femoral bones, towards the segmented femur of the patient. The most complex aspect of the mesh-to-mesh correspondence in this case lies in the fact the source (model) and the target mesh can differ largely (partial view of the femur, anatomy that lies away from the mean) which makes common correspondence approaches inefficient. In this paper, we introduce a contribution to an algorithm from the field of object recognition that produces a reliable registration. By adding the concept of global deformability in the algorithm, we are able to improve the precision of the algorithm (mean mesh-to-mesh distance improved from 2.77mm to 0.79mm) and its robustness to anatomy far off the mean (better standard deviation and Hausdorff distance) on synthetic data . The next step will be to assess it in its application field i.e. the automatic localization of knee landmarks for the design of patient-specific knee prosthesis.

Abstract. The J-Curve in the native knee as well as the femoral component’s J-Curve after total knee arthroplasty are known to have a high influence on kinematics. Furthermore, the J-Curve’s shape affects ligament strain and tension and consequently already slight changes may strongly alter knee forces and stability. To optimize current implants’ J-Curve design with regard to the population’s morphology, information about the main sources of contour variation is necessary.
In this study, a principal component analysis (PCA) was performed on the medial and lateral femoral J-Curves of 90 cadavers without history of osteoarthritis. The J-Curves’ mean shapes were further investigated by geometric parameter analysis and effect sizes were calculated for the first three principal components (PCs). In addition, a combined PCA for both sides was performed and evaluated qualitatively. The results were compared with the variation in standard implants’ J-Curve shape.
The isolated PCA of medial and lateral J-Curves resulted in PCs involving changes in contour orientation, arc length, scaling and circularity. The combined PCA of both sides resulted in PCs comprising combinations of the individual variations together with changes in relative position. In contrast, the qualitative evaluation of J-Curves from 2 different standard implant systems revealed no visible changes in shape but only changes in size.
Limitations of this study were the restriction to a 2-dimensional contour derivation and the sole consideration of the femoral contours. Nevertheless, the sagittal variability in the medial, lateral and combined femoral J-Curves should be considered in implant design.

Abstract. The anterior pelvic plane (APP) defined by both anterior superior iliac spines (ASIS) and the pubic symphysis (PS), is used as reference for cup orientation during total hip arthroplasty (THA). However, acquiring the PS and the contralateral ASIS during the intervention with the patient in lateral decubitus position, can be challenging due to the medical devices and the patient abdominal apron. The goal of this study is to find more easily accessible anatomical landmarks, useful for the APP acquisition. Thus we propose to study the variability of the pelvis anatomy in order to identify which landmarks vary with the APP. We built a statistical shape model (SSM) of the pelvis and studied the variability of APP orientation when deforming the SSM along its variation modes. We computed the APP inclination for each deformation and modeled linear relations between the APP inclination and the deformation along the variation modes. We found that the variability in APP inclination is mainly due to 3 variation modes that deform the iliac crest (IC), the posterior superior and anterior inferior iliac spines (PSIS, AIIS). Acquiring those three anatomical landmarks (IC, PSIS and AIIS) with the ipsilateral ASIS, could be a solution to determine more easily the APP for THA in lateral decubitus.

Abstract. We describe an A-mode ultrasound-based scapula tracker for pre-operative diagnosis and planning. One paper investigated this technology but did not assess its rotational accuracy, which is the critical variable. Therefore, we ran simulations and optimizations to investigate whether this technology’s rotational accuracy was adequate to make this a viable tool.
Simulations were created by converting a CT-derived scapula into an even density point cloud and positioning it in a neutral pose (our ground truth). This was reposed to a known mid-range motion, and its points were randomly sampled to mimic the data yielded by an array of single- element A-mode ultrasound transducers. This sampled dataset was registered to the ground truth using ICP, and to determine the resultant error, the resulting transformation was compared to the known transformation applied to the ground truth. The Monte Carlo method was applied to this simulation to yield varying samplings; as well, the samples were subjected to varying levels of random noise (replicating ultrasound error) to assess robustness. The effects of the number of sample points and their location were also investigated.
For 10 simulated sensors in an optimized pattern across the scapula, corrupted by random error, registration rotational errors for a sagittal plane 70◦ shoulder flexion were 0.82◦ ± 0.63◦ for posterior tilting, 0.68◦ ± 0.6◦ for upward rotation and 0.54◦ ± 0.54◦ for internal rotation.
We have successfully demonstrate that, in this simulation study, a clinically viable number of ultrasound sensors (∼ 10) could yield adequate accuracy to be diagnostically useful, if they are properly distributed across the scapula.

Abstract. The recently published DICOM-Real Time Video (DICOM-RTV) standard allows real-time synchronization of transported video and medical metadata flows. In order to demonstrate the advantages of this synchronization by design, a demonstration of a DICOM-RTV Augmented Reality (AR) based solution is proposed. This solution consists in facilitating the surgeon's gesture by monitoring the positioning of the biopsy needle under ultrasound guidance and constructing virtual and augmented reality views of the needle's trajectory. In this paper, the platform architecture and the obtained results are presented. Furthermore, a comparison between our proposed solution and a setup that do not use DICOM-RTV is provided in order to show the difference in term of precision, synchronization and delay. The obtained results validate the contribution of real-time synchronization to improve the accuracy of surgical gesture aid applications such as AR.

Abstract. Introduction
We hypothesize robotic-assisted THA can achieve precise cup positioning. The aim of this study is to figure out the accuracy of cup placement of severe DDH cases by using robotic-assisted THA
Methods
This was a prospective cohort study. The study analyzed a consecutive series of 53 hips with robotic-assisted THA between Aug 2018 and Sep 2019. Fifteen patients were DDH cases, and classified Crowe type II- 7 patients, type III- 7 patients, and type IV- 1 patient. All patients underwent robotic assisted-THA (MAKO Rio Robot, Ft. Lauderdale, FL) for osteoarthritis via the Hardinge or posterior approach in the lateral position. TTo analyze the accuracy of intraoperative navigation records for cup inclination and anteversion, we compared the intraoperative cup angles using the navigation records with the postoperative angles using postoperative CT data.
Results
Reproducibility of robotic-assisted THA for preoperative plan
In non DDH cases, the average measurement absolute error (postoperative CT-preoperative target) was 1.9 ± 2.1° (inclination) and 2.0 ± 2.5° (anteversion) (Figure 1). In DDH cases, the average measurement absolute error (postoperative CT- preoperative target) was 2.2 ± 3.1° (inclination) and 2.6 ± 2.4° (anteversion) (Figure 1). There were no differences of the average measurement absolute error between DDH and non DDH cases.
Accuracy of navigation record for cup inclination and anteversion angles
In non DDH cases, the average measurement absolute error (postoperative CT-navigation record) was 2.3 ± 1.9° (inclination) and 2.2 ± 1.5° (anteversion) (Figure 1). In DDH cases, the average measurement absolute error (postoperative CT-navigation record) was 2.0 ± 2.5° (inclination) and 2.3 ± 2.7° (anteversion) (Figure 1). There were no differences of the average measurement absolute error between DDH and non DDH cases. Scatter plots demonstrated the measurement error of both inclination and anteversion within 5° was 89.5% in non DDH cases (Figure 2a) and 80% in DDH cases (Figure 2b).
Discussion
In this study, we demonstrated that no differences of the average measurement absolute navigation record error or the absolute difference between target angle and postoperative cup alignment were found between severe DDH and non-DDH cases. Our results indicate that this robotic-assisted system has high accuracy for cup placement even in severe DDH case.

Abstract. As technology drives improvement in healthcare, the utilization trends for robotics in arthroplasty has continued to increase over the last decade. With the growth of robotics, we must determine if the proposed benefits are worth the increased cost. The purpose of this study was to evaluate inpatient post-operative and post-discharge outcomes of robotic-assisted surgery vs. conventional manual instrumentation, specifically time to discharge, discharge status and opioid consumption post- operatively. Knee Injury and osteoarthritis outcome score (KOOS) was calculated for each group as well.
After IRB approval, a retrospective chart review of 100 robotic assisted primary total knee arthroplasty (TKA) and 100 matched controls undergoing conventional TKA was performed. Baseline demographics were recorded as well as post-operative outcomes including length of stay, opioid consumption, discharge status and duration of opioid use. All patients underwent primary TKA from 2016-2018 with minimum 6- month follow-up by a single fellowship trained arthroplasty surgeon at a high-volume joint center. Exclusion criteria included < 6 month f/u, incomplete chart information, inflammatory arthritis, BMI >40.
Patients had similar pre-operative demographics including age, BMI, gender, opioid use and baseline depression rates. The robotic assisted TKA group had statistically significant decreased LOS (1.58 vs. 2.18 p< 0.001) and morphine equivalents during their hospital stay (73.52 vs. 102.50 p< 0.02). The robotic group had fewer patients at six weeks requiring opioids (37 vs. 61 p=0.001). Six month post- operative KOOS was 81.73 in the control group and 78.22 in the robotic group, (P>0.05).
Robotic-assisted TKA was associated with statistically significant decreased hospital LOS, morphine equivalents, and opioid usage at the 6-week follow up appointment. KOOS for the groups in our study were not statistically different. Although KOOS in the control group trended to be higher, our average KOOS for both cohorts was higher than the national average of 76.8.

Abstract. Patient-specific instrumentation in total knee arthroplasty (TKA), among other medical indications, requires a three-dimensional model of the bones involved. Currently, these are typically segmented from computer tomography images. Ultrasound offers a cheap as well as radiation-less imaging alternative, but suffers from a low signal-to-noise ratio as well as several other image artifacts. The interleaved partial active shape models search (IPASM) adapts a general physiological model to a set of images of a single patient, but suffers from false correspondences being soft tissue interfaces that are interpreted as bone surface. In order to counter this problem, a convolutional neural network (CNN) is applied to preprocess ultrasound images into bone confidence maps. This reduces the average surface distance error in an in-vivo evaluation by 0.7 to 1.3 mm.

Abstract. In computer assisted orthopaedic surgery, it is important to find the correct spatial lo- cation of the target in a predefined world coordinate, so that the model can be transformed accordingly onto the surgical site for surgeons’ reference. Current tracking systems mainly rely on the detection of optical markers inserted into the anatomy. The invasiveness of fixa- tion pins increases operating time and bone complications. Automatic markerless tracking is therefore preferred in practice. In this paper, we integrate an automatic RGBD-image based segmentation neural network and a fast markerless registration algorithm to achieve the markerless tracking purpose. An experimental test with a metal leg was designed. By forcing the alignment of the measured hip joint centre, the overall tracking was shown to be sub-degree in terms of orientation accuracy, which is clinically acceptable.

Abstract. Due to their sensitivity to acquisition parameters, medical images such as magnetic resonance images (MRI), Positron Emission tomography (PET) or Computed Tomography (CT) images often suffer from a kind of variability unrelated to diagnostic power, often known as the center effect (CE). This is especially true in MRI, where units are arbitrary and image values can strongly depend on subtle variations in the pulse sequences [1]. Due to the CE it is particularly difficult in various medical imaging applications to 1) pool data coming from several centers or 2) train machine learning algorithms requiring large homogeneous training sets. There is therefore a clear need for image standardization techniques aiming at reducing this effect.
Considerable improvements in image synthesis have been achieved over the recent years using (deep) machine learning. Models based on generative adversarial neural networks (GANs) now enable the generation of high definition images capable of fooling the human eye [2]. These methods are being increasingly used in medical imaging for various cross-modality (image-to-image) applications such as MR to CT synthesis [3]. However, they have been seldom used for the purpose of image standardization, i.e. for reducing the CE [4].
In this work, we explore the potential advantage of embedding a standardization step using GANs prior to knee bone tissue classification in MRI. We consider image standardization as a within-domain image synthesis problem, where our objective is to learn a mapping between a domain D constituted of heterogeneous images and a reference domain R showing one or several images of desired image characteristics.
Preliminary results suggest a beneficial impact of such a standardization step on segmentation performance.

Abstract. Using navigated analysis of bone and ligamentous status during revision total knee arthroplasty allows using frequently primary implants instead of revision ones.

Abstract. The survival rate of navigation-assisted mobile bearing total knee arthroplasty when mechanical revision was considered as the end-point wa snot impacted by the design, either central pivot or medial pivot.

Abstract. The incidence of formal soft tissues releases required in TKA surgery performed utilizing the NAVIO robotic-assisted (RA)-TKA technique was assessed and compared to a historical cohort of conventional, manually instrumented (CI) TKA’s.

Abstract. The goal of this study was to utilize the NAVIO robotic-assisted (RA)-TKA technique to assess whether a knee that is well-balanced at 0 and 90 is also well balanced in mid-flexion. Using a 3mm threshold to define soft-tissue balance, results demonstrated that 11.5% of knees studied could be expected to be unstable in the mid-flexion arc (15-75) despite being well-balanced at the static poses at 0 and 90.

Abstract. In this study, patients undergoing RA-TKA were critically assessed to understand the accuracy and precision of a simulated MR model used historically in manually instrumented TKA surgery. Using a 3mm threshold of soft-tissue laxity, knees were identified that would have been expected to require the application of a “reactive” CI-TKA surgical technique to achieve adequate soft-tissue balance.

Abstract. Fragility fracture of the sacrum has been an issue for elderly people. Research has been limited for the sacrum partly because its shape is complex with large inter-subject variation. Also, large-scale statistical analysis of its shape and density distribution has been limited mainly due to the computational load in establishment of the voxel correspondences, i.e., deformable registration.
In this study, we employed a convolutional neural network (CNN)-based deformable registration algorithm in the analysis of the sacral bone. The algorithm we employed, VoxelMorph (Dalca et al. Med Image Anal 2019), is characterized as an unsupervised algorithm where no ground truth deformation field is required. The algorithm also allows to create a conditional deformable template, which is a volume image exhibiting smallest deformation field from all samples with specific attributes (e.g., age, sex, etc.), which in short represents the “average” (or “centroid”) image among the specific age and sex group. We applied it on a database consisting of 837 CTs (149 males, 688 females, 58.14 ± 14.73 y.o.) of the pelvis region, where the sacral bone was segmented and masked. We computed the templates corresponding to ages of 20 to 80 for male and female. The templates visually illustrated reduction of the bone density with aging in both male and female. The quantitative analysis showed that the average CT value over the sacrum region was reduced from 135.73 HU to 39.88 HU for 20 y.o. template to 80 y.o. template.

Abstract. This study reports on the one-year clinical results and patient reported outcomes (PROMs) associated with a new ‘predictive balance’ tibia-cut first total knee arthroplasty (TKA) technique that uses a robotic ligament tensioner. PROMs are compared to registry data and historical results in the literature.
Five hundred and thirty-three patients were prospectively enrolled and underwent robotic TKA (mean age: 67.7±8.4; females: 320; BMI: 31.2±4.9). Pre-op, three, six and 12-month WOMAC, UCLA activity scale, and HSS-Patient satisfaction assessments were completed by 533, 352, 314, and 256 patients, respectively, and compared to WOMAC registry data from the Shared Ortech Aggregated Repository (SOAR) and to historical satisfaction reports in the literature.
Despite having equivalent baseline PROM scores, predictive balance patients had significantly higher WOMAC scores at all post-operative timepoints (p<0.001) and higher UCLA activity scale scores at 3M and 6M (p<0.013). Overall patient satisfaction in the predictive balance cohort was 91.2%, 92.4%, and 96.5% at 3M, 6M and 1YR, respectively. Average length of stay was 1.6 days (±0.8). Surgical complications in this cohort were typical of TKA.
Limitations to this study include the lack of a closely matched control group. Nonetheless, first year results are promising with improved objective measures compared to large registry databases and recently reported patient satisfaction measures.

Abstract. A well-balanced knee is a critical goal for total knee arthroplasty (TKA) across all surgical technique. Some controversy exists however about the method in which various techniques achieve soft tissue balance. Technology for balancing gaps in the tibia first approach has progressed from manual tensioners and laminar spreaders to digital tensioning tools integrated with robot-assisted navigation platforms, however there has been no comparison between femur first measured resection and tibia first techniques with this new technology. Our objective was to prospectively investigate joint balance accuracy and precision between a femur first measured resection and tibia first gap balancing technique utilizing a digital gap balancing tool.
A consecutive series cohort undergoing TKA were investigated. The cohort was divided in to 4 groups: 1) Femur First Blinded to gap data, 2) Femur First Non-Blinded to gap data, 3) Tibia first Blinded to gap data, 4) Tibia first Non-Blinded to gap data. Implanted medio-lateral (ML) gap difference was used to calculate joint balance.
94 patients were enrolled. No significant difference was found between the average ML gaps, however, the frequency of outlier balance throughout flexion, was significantly reduced in Group 4. Significant differences in the variance of the post-operative gap balance were also found, in which Group 4, reported reduced variance compared to group 3 and 2 throughout flexion and compared to group 1 at 90°.
Ideal joint balance may remain a controversial topic, nonetheless a tibia first approach with digital balancing allows surgeons to accurately achieve a desired joint balance target.

Abstract. A preoperative plan is a virtual plan that defines the implant position and orientation allowing a surgeon to prepare for surgery. However, the default preoperative plan for total knee arthroplasty proposed by the manufacturer requires changes to be made by the surgeon in more than 90% of the cases. Previous studies have shown that artificial intelligence can be used to create better preoperative plans compared to manufacturer’s default plans. However, the quality of artificial intelligence based preoperative plans has not yet been compared to surgeon approved preoperative plans. The purpose of this study is to compare default, artificial intelligence and surgeon approved preoperative plans, by having them scored on a range from 1 (totally unacceptable plan) to 5 (no corrections needed) by an experienced surgeon, while being blinded to the plan type. Through a Wilcoxon signed rank test with α=0.05, AI based preoperative plans were found to be a significant improvement upon the default plans (p-val=0.000136), while the differences in score between AI and surgeon approved preoperative plans were insignificant (p-val= 0.083). Consequently these results indicate that AI generated preoperative plans for TKA are an improvement upon current default plans, which could increase the surgeon’s planning efficiency when applied in clinical practice.

Abstract. A markerless computer aided orthopaedic platform will require a complex computer vision system to isolate and track rigid bodies used to localize a robot to a patient. Isolating rigid bodies such as bone requires accurate segmentation and this study explores using diffuse laser reflectivity to accurately classify tissue.
Lasers (Red, 650nm and IR, 850nm) intersected four material types; cartilage, ligament, muscle and metal surgical tools within a controlled cadaveric setup. Images were captured with an infrared CMOS sensor, pre-processed to isolate laser centres, and resized to test information requirements. Images for both laser types were scaled from 5x5 pixels to 30x30 pixels and trained on a convolutional neural network, GoogLeNet.
At sizes above 15x15 pixels the IR laser had a higher classification accuracy reaching 97.8% at 30x30 pixels, whereas the red laser peaked at 94.1%. It was shown as not possible to qualitatively identify materials that were not trained in the network based on their probability outputs. Further work will be done to classify multiple points in a single scene as a step toward segmenting entire surgical views for markerless CAOS systems.

Abstract. Correcting the knee joint to accommodate the pre-existing soft tissue structures is a major aim in total knee arthroplasty. 3D-to-2D registration of segmented boney geometry obtained from computer tomography (CT) scans to functionally stressed positions in X-Ray (XR) imaging can be a more accurate method of obtaining the laxity envelop. From which, a more patient specific limit for the correctability of the knee joint can be found. Supine CT scans were segmented, and 3D bone models and anatomical landmarks were registered to 2D functional stressed XR imaging. The functional position of the patient femoral and tibial bone is then used to calculate the varus and valgus extent of the patient’s knee joint laxity. 103 preoperative knees undergoing primary total knee arthroplasty identified from six different surgeons. The range of patients who are correctable back to within 3 degrees when a force is applied is 52%. 65% of patients who were within 3 degrees varus/valgus for tibia varus were correctable, while 41% of patients who were within 3 degrees varus/valgus for femoral valgus were correctable. 26% of patients were correctable when the LDFA is outside the threshold and MPTA is not, while 56% of patients were correctable when the MPTA is outside the threshold and LDFA is not. This study demonstrates the complexity of soft tissue structures of the knee joint. The main finding of this study is that correctability is more sensitive to the proximal tibial than the distal and posterior femoral articulating geometry. A lack of dependency between correctability of the knee joint and anatomical measures specific to flexion of the knee is seen. Careful consideration should be made intraoperatively when balancing the knee joint flexion gaps surrounding the soft tissue structures.

Abstract. Perceiving and making sense of the surgical scene during Total Knee Arthroplasty (TKA) surgery is crucial for building assistance and decision support systems for surgeons and their team. However, the need for large volumes of annotated and structured data for AI-based methods hinders the development of such tools. We hereby present a study on the use of transfer learning to train deep neural networks with scarce annotated data to automatically detect bony areas on live images. We provide quantitative evaluation results on in-vivo data, captured during several TKA procedures. We hope that this work will facilitate further developments of smart surgical assistance tools for orthopaedic surgery.

Abstract. Cup orientation is a challenging step in total hip arthroplasty (THA), to ensure comfort of the patient and durability of the prosthesis. The safe zone defined by Lewinnek is commonly used for cup orientation, but it is nowadays being questioned because it is not patient-specific.
We propose to define a new safe zone for cup orientation, that considers patient-specific parameters such as the pelvic tilt and the range of motion (ROM) of the leg.
We developed a software that easily computes a unique safe zone from these param- eters in different daily positions, ensuring a cup orientation without risks of prosthetic impingement.

Abstract. This study assessed the impact of adding computer-assisted orthopedic surgery (CAOS) augmentation to conventional mechanical instrumentation with regard to the enablement of both mechanical and natural alignment surgical philosophies and accuracy of postoperative alignment for total knee arthroplasty (TKA). Fifty CAOS augmented TKA cases were compared to 101 conventional cases. Data on surgical time, length of stay, and postoperative weight-bearing long leg alignment were collected. The results reported no significant impact on surgical time with incorporation of CAOS augmentation into the conventional surgical workflow, as well as a shorter length of stay and substantial decrease in alignment outliers compared to the conventional TKA cases. The study revealed the advantages of CAOS augmentation in providing a non-disruptive tool to enhance surgical accuracy and offer versatility in accommodating different surgical philosophies during TKA.

Abstract. Current surgical navigation systems offer sub-millimetric real-time localization, however they are expensive, require the use of invasive markers attached to the patient, and often add extra operation time. In this paper we propose an affordable markerless navigation approach, based on mid end depth sensors, as an alternative to answer medical applications needs in terms of accuracy and robustness. An algorithm called Fast Volumetric Reconstruction (FaVoR) implements a compute-efficient approach for real time 3D model registration based tracking, allowing computed 3D poses to be used for video scene augmentation. After early testing with a first proof-of-concept implementation, a preliminary accuracy evaluation was performed using a dynamic test bench, achieving an average 2mm registration error during tracking.

Abstract. Purpose The purpose of this study was to elucidate kinematic change according to the implant’s specific femoral rotation by using orthosensor implant with three degrees external rotation of femoral rotation rebuilt and traditional TKA implant without rebuilt of the femoral rotation .
Methods Twenty-eight patients (34 knees) underwent TKA using traditional TKA implant and 16 patients (22 knees) underwent TKA using implant with three degrees external rotation of femoral rotation. Patients were followed up for at least 1 year. Mean age of patients was 71.1 years (range, 60 to 80 years) at the time of surgery. After implantation of femur and tibial components, we applied the orthosensor system, to evaluate femoral rollback of the new artificial joint. Femoral rollback was analyzed using digitized screenshot function of orthosensor system.
Results Overall femoral tracking proportion regardless of implants was significantly higher on the medial compartment compared to that on the lateral compartment (13.3 ± 8.4% vs. 6.3 ± 5.0%, p < 0.001). Regarding femoral tracking according to each compartment, externally rotated femoral prosthesis and traditional prosthesis showed 12.1 ± 8.2% and 14.2 ± 8.6% (p = 0.371) on the medial compartment and 8.0 ± 5.8% and 5.2 ± 4.2% (p = 0.059) on the lateral compartment, respectively.
Conclusion Our study showed reverse femoral roll-back movement with higher tracking distance on the lateral compartment during TKA. externally rotated femoral prosthesis TKA system with femoral component 3-degree rebuilt showed less roll-back difference between medial and lateral compartments compared to traditional TKA system. Fortunately, both TKA systems had excellent short-term clinical outcomes without having significant difference between the two. With longer follow-up and larger cohort, the advantage and effectiveness of femoral component rotation can be elucidated in the future.

Abstract. The aim of this study was (1) to compare the clinical and radiological outcomes of robotic, navigational and conventional total knee arthroplasty with a minimum follow-up of 10 years, (2) to evaluate the survival rate, (3) and to estimate the accuracy of the three techniques by analyzing the outliers of the total knee arthroplasty (TKA) patients.
Methods From January 1992 to December 2008, We evaluated 515 knees who underwent robotic TKA, navigational TKA, or conventional TKA with minimum follow-up of 10 years. Finally, this study including 92 patients (103 knees) who underwent robotic total knee arthroplasty using ROBODOC® , 197 patients (225 knees) who underwent navigational total knee arthroplasty using Orthopilot, and 175 patients (187 knees) who underwent conventional total knee arthroplasty. Hospital for special surgery(HSS) score, Knee Society Score(KSS), Western Ontario and McMaster Universities (WOMAC), and Range of Motion(ROM) were used for clinical evaluation. Mechanical alignment, implant radiological measurements and outliers were analyzed for radiological results. Complication related with surgery was also evaluated. Data were analyzed using SPSS Version 24.0 software and descriptive statistics, paired t‑test, one‑way ANOVA, and Wilcoxon tests. Kaplan-Meier survival analysis was performed for survival rate.
Results All clinical assessments including HSS, KSS, WOMAC, and ROM at final follow-up showed excellent improvements in three groups (all, p<0.05), without any significant difference between the groups (p>0.05). In radiologic data, sagittal inclination of femoral implant in robotic group showed better result than another groups with significant difference(p<0.05). The cumulative survival rate was 94.8% in the robotic group, 96.2% in the navigation group, and 92.4% in the conventional TKA group with excellent survival (p=0.563). Complication rate associated surgery was 5.2 % in the robotic group, 5.3% in the navigation group, and 8% in the conventional group
Conclusion Our study showed excellent survival with robotic, navigation and conventional TKA and similar clinical outcomes at long-term follow-up. However, in terms of radiological outcome robotic TKA showed accurate position of femoral component. With longer follow-up and larger cohort, the accuracy and effectiveness of robotic TKA on implant survival rate can be elucidated in the future.

Abstract. This abstract addresses the problem of localizing surgical instruments during orthopaedic surgeries. Compared to usual approaches based on surgical navigation with markers, we propose here a novel method that estimates the 6-DoF pose of surgical instruments without specific markers using a depth camera. The goal of this paper is to compare, on real data, the registration precision of an algorithm called Point Pair Features (PPF) according to consumer depth cameras available on the market. Experimental validation using sawbones has been conducted and 8 cameras have been tested in realistic clinical environment. The Kinect Azure reports the best precision with a registration error of 1.13mm ± 1.00mm.

Abstract. The goal of total knee arthroplasty (TKA) is to position the prosthesis with a bal- anced soft tissue envelope throughout the flexion cycle. Determining a desirable amount of ligament tension is crucial as instability is a leading cause for revision surgery. This is challenging due to the subject specific, and non-linear, nature of ligament mechanical properties. This study aims to characterize ligament stiffness profiles intraoperatively and identify the stiffness transition point (STP) using a ligament balancing robot. The study will also identify how the surgeon selected joint force relates to the stiffness transition point.
45 patients were reviewed. After the proximal tibial resection, intraoperative assessment of the medial and lateral ligaments was performed via a load cycle ramped up from 50N to 120N, then down from 120N to 50N. This was performed at extension (10◦) and in flexion (90◦). Force and displacement data were processed to determine the stiffness profiles for the medial and lateral soft tissue envelope. A bilinear fit model was used to determine the slopes and STP.
The average STP was between 83N and 90N, varying widely with standard deviations approaching 14N. The median joint tension selected by the surgeon was 80N. On average the selected joint force was 3N to 8.9N below the STP.
The medial compartment behaved similarly in extension and flexion. The lateral com- partment had higher stiffness in extension than in flexion. Across all loading conditions, the down-cycle data was more consistent than the up-cycle data.
The STP is a proposed target for ligament tension as this theoretically avoids the high stiffness regime. The results show the joint tension selected by the surgeons, based on their experience and intraoperative feel, is similar to the STP. Due to the high patient variability observed in the STP, using a patient specific method to determine the joint tension is recommended. Future work will investigate how joint tension relative to the STP affects patient outcomes. This will provide insight into optimizing joint tension during TKA.

Abstract. Surgical navigation can be used for complex orthopaedic procedures, such as iliosacral screw fix- ations, to achieve accurate and efficient results [11]. Although there have been studies studying the impact of navigation systems on surgical outcomes [6, 3], we are not aware of any studies that have quantified the effect of how information regarding the surgical navigation scene is displayed to surgeons on conventional monitors. However, the display of information can have a measurable effect on both accuracy and time required to perform the navigated surgery, as the surgical scene can be presented in many different formats [9]. Optimizing surgical accuracy potentially helps improves patient safety by reducing screw malplacement [11], while optimiz- ing time efficiency reduces opportunity cost [1]. Therefore, we designed a study to determine the optimal visualizations for performing navigated pelvic screw insertions. The findings of this study can be used to more systematically design visualization components of a navigation system.

Abstract. In pedicle screws placement using a free-hand technique or a fluoroscopic guided technique the main difficulties are facing to the bone morphology (i.e in deformity cases) and it could be easily reproduced in a patient’s specific spine simulator (we can choose the case). The aim of this work is to evaluate the use of 3D printed patient- specific models (3D printing) not only as a surgical planning tool but also as a surgical training tool in spine surgery and in particular in pedicle screws placement. The manufacturing of patient-specific physical replica involves the elaboration of CT dataset and rapid prototyping techniques. . Five resident surgeons were involved in different training sessions on simulators

. To evaluate the exact screws position we
performed a CT evaluation of each instrumented simulators
. Statistical analysis was conducted using SPSS software. A total of 120 pedicle screws were positioned, 90 screws were well-positioned and 30 screws were bad-positioned. There were a significant difference (p = 0.000008) between the bad-positioning screw rate of the “senior” resident (13/72) and those of “young” participants (17/48). Timeline analysis of pedicle instrumentation training showed the presence of a learning effect, with a lower error rate in the latest session (p=000001). We believe that the use of patient- specific surgical simulators, especially for those surgical tasks in which the complexity is mainly linked to the spine morphology (i.e. deformity), may represent a valid alternative to the use of cadavers that generally present a standard or otherwise poorly predictable anatomy.

Abstract. Objectives: Applying the correct amount of collateral ligaments tension in the knees during surgery is a prerequisite to restore normal kinematics after TKA. It is well known that a low value of ligament tension could lead to an instable joint while a higher tension could induce over-tensioning and problems at later follow-up. In this study, an experimental cadaveric activity was performed to measure the minimum tension required to achieve stability in the knee joint.
Methods: 10 cadaveric knee specimens were investigated in this study. The femur and tibia were fixed with polyurethane foam in specific designed fixtures and clamped to a loading frame.
Increasing displacement was applied to the femoral clamp and the relative force was measured by a loading-frame machine up to the stability of the joint, determined by a decrease in the derivate of the force/displacement trend followed by a plateau.
The force span between the slack region and the plateau was considered as the tension required to stabilize the joint.
This methodology was applied for joints with intact cruciate ligaments, after ACL resection and after further PCL resection, to simulate the knee behavior prior a CR and a PS implant.
The test was performed at 0, 30, 60 and 90° of flexion. Each configuration was analyzed three times for the sake of repeatability.

Results and Conclusion: Results demonstrated that an overall tension of 41.2N (range 30.0-48.0 N) is sufficient to reach stability in a native knee with intact cruciate ligaments. Similar values appear to be sufficient also in an ACL resected knee (average 45.6, range 41.2-50.0 N), while higher tension is required (average 58.6N, range 40.0-77.0 N) were necessary in the case of PCL retention. Moreover, in this configuration, the tension required for stabilization was slighter higher at 30 and 60° of flexion compared to the one required at 0 and 90° of flexion.
The results are in agreement to the ones found by other recent experimental study [Manning et al 2018 (KSSTA)] and shown that the tension necessary to stabilize a knee joint in different ligament conditions is way lower than the ones usually applied via tensioners nowadays.
To reach functional stability, surgeons need to consider such results intraoperatively to avoid laxity, mid-flexion instability or ligament over-tension.

Abstract. Introduction
Worldwide, the number of TKA implants is increasing. Even if registry demonstrate that TKA as high satisfaction rate, there are still between 15 and 20% dissatisfied patients.
Materials and methods
The proper soft tissue balancing is one of the most discussed topics of the last years. We initiated a study using an electronic device („dynamic ligament balancing sensorplate“) to compare the benefit of the measurement of ligament tension, space and position in comparison to a conventional surgical procedure. Beside that, we followed the concept of functional stability, which tells us, that a tension of 40N in total is sufficient to reach proper (functional) joint stability.
This control pilot study was set up as a single surgeon, single center study and consists of 25 patients treated by the use of the sensorplate and a control group of 25 patients, treated in a conventional setup.
We used the following scores for evaluation: OKS, AKSS and FJS, preoperatively and during the FU examinations (postoperative, 6 weeks, 3 months, 6 months and 1 year)
Beside scoring, clinical examination and routine x-ray we performed an EMG testing at all FU dates.
Results
The study was performed between January 2017 and May 2019. The mean age of the patients was in average 72 years, 66 % female and 34 % male. After 1 year, results demonstrate a clear difference in the development of the postoperative situation between dynamic balanced TKA and the control one.
So, the use of such electronic device improving the soft tissue envelope stability, enable a significantly better patient FU, especially in terms of OKS
Discussion
The DLB system is a new option to value and improve the soft tissue envelope tensioning during the surgical TKA operation. It allows to measure ligament tension, slope and joint space all over the entire ROM.
Conclusion
Using an electronic device for measurement is an advanced option to improve patient satisfaction after tka. Like the studies of other existing devices have shown before there is a massive change in the kinematic behaviour of the muscular abilities by using these tools for a better soft tissue balance. The DLB system is another option by showing 3 different measurement results (tension, distance and joint angle) to adapt the implantation procedure to the individual situation of the patient.

Abstract. Adequate deltoid and rotator cuff lengthening in total shoulder arthroplasty (TSA) is crucial to maximize the postoperative functional outcome and to avoid complications. Hence surgeons and patients could benefit from including muscle length information in preoperative planning software.
Although different methods have been introduced to automatically indicate patient-specific muscle attachment and wrapping points, the definition of a fast and accurate workflow is still a challenge, due to the large variability in bone shapes. Therefore, the goal of this study is to develop and evaluate the accuracy of a novel method to automatically estimate muscle elongation, based on a statistical shape modelling (SSM) approach.
Firstly, humerus and scapula SSMs were used to automatically indicate the attachment points of the main shoulder muscles: subscapularis, supraspinatus, infraspinatus, teres minor and deltoid. Secondly, a wrapping algorithm was applied to identify the points where muscles wrap around bones or potential implants. Finally, the accuracy of the automatically indicated landmarks and its effect on the muscle elongation were evaluated by comparing the manually indicated landmarks with the landmarks identified through the SSM for a set of 40 healthy shoulder CT-scans.
The low errors on elongation values suggest that the presented automated workflow is a promising tool for allowing surgeons to evaluate patient-specific muscle elongations during preoperative planning. Although the evaluation was limited to healthy joints, this method allows to easily process large datasets and to potentially find a correlation between muscle elongations and postoperative outcome.

Abstract. Soft tissue releases are often required to correct deformity and achieve balance in total knee arthroplasty (TKA). However, releasing soft tissues can be subjective, highly variable and is perceived as an ‘art’ in TKA. The objective of this study was to compare the rate of soft tissue release required to achieve a balanced knee in tibial-first gap- balancing versus conventional, measured resection TKA, and its effect on outcomes.
Soft tissue releases were documented and reviewed in 1256 robotic-assisted gap- balancing and 85 robotic-assisted measured-resection TKAs. Knees were stratified by coronal deformity (varus: >2° varus; valgus: >2° valgus). Rates of releases were compared between the two groups and literature. A subset of these patients were also enrolled in a prospective study. KOOS outcomes were captured pre-operatively and at 6M post TKA.
The frequency of soft tissue release was significantly lower in the robotic gap- balancing group, with 21% of knees requiring release versus 40% (p=0.001) in the robotic measured resection group and 67% (p<0.001) for conventional measured resection. Pre-operative KOOS scores were similar between groups, however 6M scores showed a significant improvement in QOL, Sports and Symptoms scores in knees not released.
Robotic assisted TKA with predictive gap balancing was found to reduce the number of releases across all coronal angles compared to conventional instruments. Furthermore, performing a soft tissue release rather than bone resection to achieve balance, correlated with worse outcomes. Further research is required to understand when imbalance should be corrected with bone resection adjustment versus soft tissue release.

Abstract. A cohort of 84 patients underwent Total Knee Replacement surgery using Patient Specific Instrumentation fitted with an optical tracker that was monitored by a Computer Assisted Surgery system. The CAS system was low cost with small footprint in the operating theatre. The hip centre was collected and four other landmarks were recorded as rotational measures. The CAS system then reported the deviation in PSI placement against the targeted PSI placement, the surgeon then made a judgement whether to replace the guide. Post-operative analysis was done to determine the achieved alignment of the component and compared to the targeted alignment. From 45 results available for analysis the study found that the mean of rotational measures over the femur and tibia were found to be within ±30 of the targeted alignment, except for femur sagittal alignment. When a single outlier was removed from the femur sagittal alignment measures, the mean fell below ±30 of the targeted alignment. Distal femoral condyle resection measures fell below a mean of 1mm and posterior femoral condyle measures fell below a mean of 2mm. Lateral proximal tibial resection fell below a mean of 2.5mm as did medial proximal tibial resection when two outliers were removed. This shows that a CAS system incorporating the low cost, quicker time, and smaller footprint benefits of PSI with the accuracy of traditional navigation may be a feasible device.

Abstract. Computer-Assisted Orthopaedic Surgery (CAOS) is now becoming more prevalent, especially in knee arthroplasty. CAOS systems have the potential to improve the accuracy and repeatability of surgical procedures by means of digital preoperative planning and intraoperative tracking of the patient and surgical instruments.
One area where the accuracy and repeatability of computer-assisted interventions could prove especially beneficial is the treatment of osteochondral defects (OCD). OCDs represent a common problem in the patient population, and are often a cause of pain and discomfort. The use of synthetic implants is a valid option for patients who cannot be treated with regenerative methods, but the outcome can be negatively impacted by incorrect positioning of the implant and lack of congruency with the surrounding anatomy.
In this paper, we present a novel computer-assisted surgical workflow for the treatment of osteochondral defects. The software we developed automatically selects the implant that most closely matches the patient’s anatomy and computes the best pose. By combining this software with the existing capabilities of the Navio™ surgical system (Smith & Nephew inc.), we were able to create a complete workflow that incorporates both surgical planning and assisted bone preparation.
Our preliminary testing on plastic bone models was successful and demonstrated that the workflow can be used to select and position an appropriate implant for a given defect.

Abstract. For preoperative simulations of hip range of motion to be useful in predicting complications after total hip arthroplasty (THA), the factors that could affect post-operative function must be considered including, but not limited to, bony impingement, pelvic position, and implanted vs. planned differences. This study retrospectively simulates ranges of motion to prosthetic and bony impingement of THA patients with known planned and implanted component positions and pelvic tilt to determine the factors and needs to accurately simulate range of motion preoperatively.
Twenty-two (22) anterolateral, cementless total hip arthroplasties were performed using robotic-arm assisted technology which allowed capture of the implanted stem version and position in addition to robotic-assisted cup placement to plan. With the known implanted positions and preoperative 3-dimentional (3D) bone models, six (6) hip maneuvers were virtually simulated in custom software. Correlations were evaluated between planned and implanted component positions, pelvic tilt, ranges of motion, and patient-reported outcomes.
Average ranges of motion to impingement were similar to those of previous simulation and navigation studies. Supine tilt varied from -10 ̊ (posterior) to 15 ̊ (anterior) with an average of 3.4±6.6 ̊. Very little correlation was seen between native or planned stem version and implanted stem version. Correlations were seen between some maneuvers such as internal rotation (IR) at 90 degrees flexion (F) (IR@90F) and combined component version and pelvic tilt. Bony impingement occurred during IR@90F in 9 of the 22cases. Pelvic tilt assessment, bony impingement detection, better prediction of implanted component position or the ability to execute a plan, such as robotically, would all provide a more accurate pre-operative simulation of the post-operative patient’s function.

Abstract. Dynamic knee computer simulations are a promising surgical planning option in TKA, allowing the impact of plan alterations on joint dynamics to be analysed prior to surgery. Previously, the dynamic results of our simulation have been shown to correlate with outcome; here we show validation of its use in pre-operative planning.
A database of TKA Patients undergoing surgery from 1-Jan-2014 operated on by 9 surgeons, who received a pre-operative and post-operative CT were assessed. A musculoskeletal computational model with similar boundary conditions to the Oxford Knee Rig was used to simulate post-TKA knee dynamics using Adams MSC software (Newport, CA). In addition, a set of pre-operative simulations were generated covering positional variations. The Dynamic Knee Score (DKS), a predictive algorithm machine learned from KOOS scored postoperative cases to predict outcome in preoperative planning was applied to all simulations.
Patients were split into groups depending on whether the ‘post-operative achieved position’ was the ‘best’ of the preoperative modelled options in terms of simulated DKS score or not. These results were compared with 12 month postoperative KOOS scores. Cases where the best plan was followed had better outcome results. A relationship was shown with the KOOS Pain subscore, with the portion of patients below a KOOS Pain score of 70 dropping to 11% from 16% (p=0.030) when the best plan was followed.
This study shows significant relationships between selection of patient specific kinematically optimal surgical plan and outcome. Such tools will play an important role in future patient specific decision making.

Abstract. Currently, patient selection for Total Knee Arthroplasties (TKA) is surgeon specific. A combination of patient reported symptoms, clinical examination findings, and radiological criteria are combined in an idiosyncratic fashion based on the surgeon’s individual clinical experiences. Predictive models offer an alternative by providing more detailed information on expected patient outcome. This study presents validation relative to surgeon predictions of such a predictive model, the Patient Expectation Management (PEM) tool.
A cohort of patients undertook a survey covering the KOOS questionnaire and a number of other questions pertaining to comorbidities prior to their consultation. From this survey, a prediction of final state and assessment of current state was generated. Prior to seeing the prediction but after consulting the patient, surgeons were asked to score out of 100 a) their understanding of the patient’s current pain state and b) their prediction of the patient’s pain level following surgery. 35 of the patients were selected for TKA surgery and have gone on to have 12 month Knee Osteoarthritis and Outcome Scores (KOOS) captured.
The predicted change in the PEM predicted score (preop to postop difference) had a relatively high correlation with the actual KOOS pain improvement achieved (r=0.71, p<0.001), compared to no significant correlation for the surgeon prediction (r=0.24, p=0.20). Significant correlations also existed for changes in KOOS symptoms score (r=0.70, p<0.001) and KOOS Activities of Daily Living (ADL) score (r=0.42, p=0.02).
This study showed that, compared to a set of surgeon predictions of outcome following a consultation with patients, a predictive analytics tool was able to outperform in terms of predicting the improvement patients are likely to report following TKA.

Abstract. Femoral head coverage in patients with hip dysplasia (DDH) is often quantified using 2D parameters, including the lateral center edge angle (LCEA) and anterior center edge angle (ACEA). However, only moderate correlations have been observed between the 2D coverage and 3D coverage. The purposes of this study were to: 1) quantify the change in 3D head coverage after curved periacetabular osteotomy (CPO), and 2) analyze the relationship between 2D and 3D head coverage preoperatively and postoperatively. Forty-three hips of 39 female cases (age: 37±10 years) who underwent CPO were analyzed. 2D coverage was quantified using the LCEA and ACEA from CT images. 3D coverage was quantified in the anterior, superior, posterior, inferior head regions. 3D measurements were performed both pre- and post-operatively and were correlated to the measurements of 2D coverage to study interactions. Preoperative 3D percent coverage in each head region was 17.8±6.7%, 36.2±7.7%, 57.6±10.2%, and 15.3±6.4% for the anterior, superior, posterior, and inferior region, respectively. Postoperatively, 3D coverage in the anterior and superior regions increased to 23.4% and 53.7%, respectively while a significant decrease was found for the posterior and inferior regions (all p<0.01). When 3D and 2D coverage was correlated, significant positive correlation was found between the 3D superior coverage and the LCEA both preoperatively (r=0.72, p<0.01) and postoperatively (r=0.67, p<0.01). However, no correlation was found between the 3D anterior coverage and the ACEA, which became significant in the postoperative period (r=0.69, p<0.01). Results indicate that preoperative anterior coverage for patients with DDH should be evaluated three-dimensionally.

Abstract. Background: Spinal stiffness has been shown to increase risk of dislocation due to impingement and instability. Increasing anteversion of the acetabular component has been suggested to prevent dislocation, but little has been discussed in terms of femoral or global offset restoration. The purpose of this study is to quantify dislocation rates after primary THA using standard versus high offset femoral components and to determine how differences in offset affect impingement-free range of motion in a stiff spine cohort using a novel impingement model. Methods: 12,365 patients undergoing THA from 2016-2018 were retrospectively reviewed to determine dislocation rates and utilization of standard versus high offset stems. For 50 consecutive patients with spinal stiffness, a CT-based computer software impingement modeling system assessed bony or prosthetic impingement during simulated range of motion. The model was run 5 times for each patient with varying offsets. Range of motion was simulated in each scenario to determine the degree at which impingement occurred. Results: There were 51 dislocations for a 0.41% dislocation rate. Total utilization of high offset stems in the entire cohort was 49%. Of those patients who sustained a dislocation, 49 (96%) utilized a standard offset stem. The impingement modeling demonstrated 5 degrees of added range of motion until impingement for every 1mm offset increase. Conclusion: In the impingement model, high offset stems facilitated greater ROM before bony impingement and resulted in lower dislocation rates. In the setting of high-risk THA due to spinal stiffness, surgeons should consider the use of high-offset stems and pay attention to offset restoration.

Abstract. Current surgical robotic systems consist either of a large serial arm, resulting in higher risks due to their high inertia and no inherent limitations of the working space, or they are bone-mounted, adding substantial additional task steps to the surgical workflow.
The robot presented in this paper has a handy and lightweight design and can be easily held by the surgeon. No rigid fixation to the bone or a cart is necessary. A high-speed tracking camera together with a fast control system ensure the accurate positioning of the burring tool, while automatically compensating for movements of the surgeon or the patient’s bone.
To evaluate the motion compensation capabilities of the developed robot, an experiment was conducted in which movements of the patient were simulated on the robot’s real time control system and the actual robot had to follow those virtual movements. The positioning error, measured with a tracking camera, was recorded for different velocities of the disturbing movement.
A linear relation between the positioning error and the speed of the disturbing motion could be discovered, with a slope of 24 ms and an offset of 0.044 mm.
The slope can be interpreted as the latency of the robotic system while following a moving target. Therefore, with a measured latency of 24 ms, the developed robotic system should easily be able to compensate for the patient’s breathing as well as the tremor of the operator, whereas the latter can reach frequencies of up to 12 Hz according to literature.

Abstract. A poorly balanced, unstable or stiff joint is a leading cause of dissatisfaction and revision after surgery. The quantitative definition of a well-balanced joint, however, remains a source of controversy. This study investigates joint gaps measured by a digitally-controlled ligament tensioning device throughout flexion and its effect on post- operative outcome.
Surgeries were performed using the OMNIBotics robot-assisted TKA platform and BalanceBot device. Joint gaps were recorded by the BalanceBot throughout flexion during trialing. Knee Injury and Osteoarthritis Outcome Score (KOOS) was collected at 1-year post-op. Correlations between joint gaps and KOOS outcome were investigated.
Knees that reported: a medial gap ≤ lateral gap in extension (p = 0.007, median ∆ = 8.3); an average joint gap of between 1 mm tighter and 1 mm looser than the final insert thickness in midflexion (p = 0.006, median ∆ = 5.5); and an imbalance of less than 1.5 mm in flexion (p = 0.012, median ∆ = 2.8) reported significantly improved pain scores. Patients which satisfied both extension and flexion criteria, or midflexion and flexion criteria reported improved outcomes compared to those which satisfied only one or neither criterion (p = 0.0002, median ∆ = 9.7, p = 0.0019, median ∆ = 8.4 respectively).
KOOS Pain scores correlated with joint gap measurements across all flexion angles investigated. Combining joint gap windows, subgroups of patients were found that reported a change in the median KOOS Pain outcome beyond the minimally clinically important difference.

Abstract. Achieving a balanced knee is a critical aspect of Total Knee Arthroplasty (TKA). Coronal and axial boundaries for femoral component placement to achieve balance however, are not well defined. Our aim is to investigate the effect of femoral component and long leg coronal and axial alignment on patient outcomes when using a tibia-first gap balancing technique.
All surgeries were performed using the OMNIBotics robot-assisted TKA platform and BalanceBot device. A total of 197 patients were prospectively enrolled into this study and received TKA surgery using the OMNIBotics platform and completed 1-year KOOS outcome scores. Femoral component and tibiofemoral alignment were categorized as inliers or outliers in the coronal and axial planes. Knee Injury and Osteoarthritis Outcome Score (KOOS), and University of California at Los Angeles Activity Scale (UCLA) was collected at 1-year post-op.
No significant differences were found between the KOOS subscores or UCLA outcome and femoral coronal or tibiofemoral coronal and axial alignment. Significant differences were found between the KOOS pain and sports sub-scores and femoral axial alignment (∆ = 5.4, p = 0.007, ∆ = 8.3, p = 0.03 respectively), in which outlier femoral rotation reported higher scores.
Component alignment limits for improved survival and patient outcomes are a source of ongoing debate. The data presented here indicates that when utilizing a tibia-first gap balancing technique, small deviations outside of traditional ±3°alignment boundaries did not negatively affect KOOS or UCLA outcomes, indicating balance may have a stronger link to patient outcome than alignment.

Abstract. Joint balance in total knee arthroplasty (TKA) has traditionally focused on achieving a tight symmetric extension gap and rectangular or trapezoidal gaps in flexion. This study sought to investigate the effect of femoral and tibial coronal rotation and femoral axial rotation on midflexion coronal joint balance and patient outcomes.
A prospective multi-center study was performed with a mixture of tibia-first gap-balancing and femur-first approaches were performed using the Corin OMNIBotics robot-assisted TKA platform with APEX implant components. Coronal and axial femoral and tibial resections were recorded by the platform. Medial and lateral joint gaps were recorded while applying a computer-controlled load to the joint throughout flexion during trialing using the Corin BalanceBot device. In addition, 1-year Knee Injury and Osteoarthritis Outcome Score (KOOS) and PROMIS-10 global health scores were collected.
231 surgeries were identified: 66.9±8.1 years, 31.4±4.8 kg/m2 and 57% female (121) with a mean pre-operative HKA angle of 4.5±5.2° varus. A significant correlation was found between the medio-lateral (ML) joint gap difference in midflexion and both extension and flexion joint line (p=0.003, r2=-0.20, p=0.001, r2=-0.22, respectively). A significant correlation was found between midflexion ML imbalance and KOOS stiffness questions at 3 M and 6 M post-op (r2=-0.15, p=0.036, r2=-0.18, p=0.013), in which a more balanced knee correlated with improved outcomes.
Treating flexion and extension joint balance in isolation may not capture the effect on midflexion laxity. Component placement should take in to account the effect on joint gaps throughout flexion to target optimal joint balance.

Abstract. Laser Powder Bed Fusion (LPBF) is an additive manufacturing process, which enables the generation of complex geometries such as lattice structures, using metallic powder. Lattice structures are being used increasingly in medical technology to adapt the stiffness of individualized implants which can lead to faster bone healing. Lattice structures are also used to adjust the contact surface between the bone and implant to adapt the osseointegrative behavior. The goal of this work is to create lattice structures with local adaption of the stiffness (modulus of elasticity) for individual vertebral body replacement implants and their automated design based on patient data.
To form the lattice structure a diamond cell type is used, which is common in medical technology. For the later adaptation of the bone stiffness, the stiffness of the lattice structure with different strut diameters are determined. The calculation of the stiffness is done by numerical simulations using Finite Element Methods (FEM). The simulations are validated with tensile and compression tests. Finally, the automated design of the implants is carried out with an in-house generated tool to adjust the strut diameters based on the bone density from patient data.
Parts of this work have been funded by the German ministry of education and research (BMBF) under grant number 13GW0116.

Abstract. Dislocation of the bearing occurs in 1 - 6% of Oxford Domed Lateral (ODL) mobile bearing unicompartmental knee replacements (UKRs). Dislocations occur in flexion as the lateral ligaments are lax in this position allowing the knee to distract. Anterior and posterior dislocations are rare: clinically, their dislocation rates are acceptable. Most dislocations tend to occur medially, with the bearing sitting on top of the tibial wall. Using robotics path planning algorithms and a modified Open Motion Planning Library (OMPL) Graphical User Interface (GUI), a dislocation analysis tool was developed to assess the minimum amount of vertical distraction of the femoral component relative to the tibial component required for the mobile bearing to dislocate. In the tool, the Rapidly- exploring Random Trees (RRT) algorithm was applied to the mobile bearing, which enabled autonomous movement of the bearing from a non-dislocated to a dislocated position. Testing increased the relative distance between the femoral component and the tibial component: vertically (2-6 mm) and mediolaterally (0-4 mm) in 0.25 mm increments resulting in a total of 289 configurations. For each configuration, the tool assessed whether mobile bearing dislocation was possible (either medially, laterally, anteriorly or posteriorly). For each mediolateral translation distance, the minimum vertical distraction required for dislocation was recorded. To validate the tool, dislocation results were compared to measurements taken using a custom-built mechanical rig. The minimum amount of distraction required for medial dislocation was similar for the dislocation analysis tool (3.75 to 4.75 mm) as compared to a custom-built mechanical rig (2.5 to 4 mm). The amount of distraction for a medial dislocation was much smaller than that for an anterior or posterior dislocation (6 to 6.25 mm). This explains why medial dislocations are more common. Future work will use this tool to inform implant design, with the aim to reduce the risk of medial dislocation to match that of anterior/posterior dislocation, which is clinically acceptable.

Abstract. In ultrasound (US)-based computer-assisted orthopedic surgery (CAOS), accurate and robust intra-operative registration in real-time is vital in securing the reliable outcomes for surgical image guidance. For this purpose, we focus on developing a hierarchical registration method, using reinforcement learning (RL), for 3-D registration of pre-operative computed tomography (CT) data to intra-operative US. In the RL-based registration procedure, we proposed a supervised Q-learning framework for learning the sequence of motion action to achieve the optimal alignment. Within the approach, the agent was modeled using PointNet++ with the mis-aligned point set from US and CT as the input, and the next optimal action as the output. Evaluation studies achieved average target registration error (TRE) of 3.82 mm with success rate of 92.7% and an average time of 8.36 seconds. We achieve 57.1% improvement in success rate over state of the art.

Abstract. We present a fully automatic method of segmenting and landmarking hip CT images for planning of Total Hip Arthroplasty (THA). Our method consists of two stages, i.e., the segmentation stage and the landmarking stage. At the segmentation stage, a multi-atlas segmentation constrained graph method is employed to fully automatically segment both the pelvis and the bilateral proximal femurs from the input CT data. The segmentation stage is followed by the landmarking stage, where a set of pre-defined landmarks are transferred from generic models of the associated hip structures to the input CT space via non-rigid registrations in order to compute a set of functional parameters that are relevant to planning of THA. Evaluated on 20 hip patients, we computed both the segmentation accuracy and the landmarking accuracy. An average segmentation error of 0.38 ± 0.25 mm and 0.49 ± 0.22 mm was found for the hemi-pelvis and for the proximal femurs, respectively. For 3D landmarking, a mean error of 1.58 ± 0.87 mm and 0.46 ± 0.39 mm was found for the acetabular rim center and the acetabular rim radius, respectively; a mean error of 0.74±0.45o was found for the orientation of the anterior pelvic plane; and a mean error of 3.14 ± 1.90 mm and 2.04 ± 1.61 mm was found for the femoral head center and the femoral offset, respectively.