ABSTRACT. Introduction In orthopedics, surgical robots rely on preoperative planning derived from diagnostic imaging to establish a global action roadmap and guide high-level robotic control. While these plans accurately represent the patient's anatomy prior to surgery, they offer only a static and simplified representation of the reality. Intraoperative factors such as patient motion, unexpected anatomical variations, and complications significantly influence the surgery, requiring surgeons to continually reassess and adapt their initial plans. The development of a closed-loop robotic system capable of optimally adapting to intraoperative conditions remains challenging due to the insufficient integration of real-time intraoperative data. The European Union project FAROS, short for “Functionally Accurate Robotic Spine Surgery,” aims to achieve this goal by introducing a significant departure from rigid preoperative planning paradigms. Methods FAROS is founded on the scientific hypothesis that integrating non-visual intraoperative sensing technologies enhances real-time robotic decision-making and autonomy in spine surgery. Two relevant clinical use cases were investigated, namely autonomous robotic pedicle screw placement (PSP) and robot-assisted endoscopic discectomy (ELD) both based on a custom-developed multi-arm robotic platform based on Kuka’s LBR Med 7/14 (KUKA, Augsburg, Germany). Various sensor modalities were explored to enhance real-time robotic perception, including robotic ultrasound (US) for intraoperative 3D reconstruction of spinal anatomy and real-time breach detection using vibroacoustic and impedance sensing. A hyperspectral robotic endoscope was developed to improve tissue differentiation, complemented by novel robotic control algorithms for robust force control and seamless human-robot interaction. Results The final robotic prototype for autonomous PSP consisted of a single robotic arm integrating US-based anatomy reconstruction and registration, optical breathing compensation, a robotic drill end effector, and conductivity sensors for real-time breach detection. Validation was conducted through robotic drilling of lumbar PSP trajectories on an ex-vivo human torso and an in-vivo porcine model. In ex-vivo, a 3D accuracy evaluation against the preoperative plan showed a 3D translational error of 3.8 ± 2.11 mm and a 3D rotational error of 4.2 ± 2.2°. The system successfully placed 77% of screws with a grade “A” according to the Gertzbein-Robin classification. The breach detection system achieved a 71% success rate in detecting breaches. However, in-vivo experiments presented additional challenges, including reduced accuracy due to registration issues, drill slippage, and soft tissue tension. The FAROS robotic ELD solution was qualitatively evaluated ex-vivo by an experienced spine surgeon. It incorporates radiation-free ultrasound guidance and a robotic endoscope holder with multiple interaction modes.

ABSTRACT. Acoustic signals have been utilized by medical professionals for centuries, for example for the diagnostic assessment of human body sounds using a stethoscope, but are rarely used in conventional computer aided diagnostics and surgery. However, acoustic signals have great potential for the development of novel multimodal sensing solutions for medical applications and can provide solutions for problems where conventional systems, such as surgical navigation or medical imaging reach their limits. Our research in the field of acoustic sensing shows that the combination of a highly sensitive sensor technology, advanced signal processing, and powerful cutting-edge analysis methods based on Deep Learning enables the utilization of acoustic signals for the design of easy-to-integrate, non-invasive, radiation-free, and low-cost multimodal sensing systems in computer aided medicine. Our contributions include solutions to analyze semantic information for unmet clinical problems in diagnostics and surgical interventions, such as implant loosening detection for pedicle screws in spinal fusion surgery, drill breakthrough detection in orthopedic bone drilling, and the identification of the optimal insertion endpoint for the femoral stem component in Total Hip Arthroplasty using structure-borne hammer blow sounds. All proposed systems were thoroughly evaluated in extensive and realistic experimental setups.

Apart from semantic information and the detection of acoustic events in surgery, context awareness and scene understanding are of great interest for the development of the next generation of intelligent systems in computer-aided and robotic surgery which require an understanding of the spatial relationships of a surgical scene. To extend the concept of analyzing semantic information based on acoustic activity, we introduce the novel concept of Sound Source Localization for surgery which can be employed to reveal not only the semantics, but also the location of acoustic activity in the surgical field, therefore providing detailed insights into the interactions of surgical staff with the patient and medical equipment. This concept be an important step towards enhancing context understanding capabilities of intelligent systems in the operating rooms of the future.

Our results proof that automated decision and support systems based on acoustic sensing have great potential for the development of new multimodal sensing paradigms for a wide range of applications in medical diagnostics, interventions, and the analysis of surgical workflows. In this context, acoustic sensing systems can be utilized to complement existing computer aided surgery, surgical guidance, and decision support systems and provide information beyond the limits of established methods such as surgical navigation systems or medical imaging.

ABSTRACT. Total hip arthroplasty (THA) is among the most common surgeries for hip osteoarthritis. Besides the conventional manual technique (mTHA), alternatives such as computer-assisted fluoroscopic navigation (cTHA) and robotic-assisted solutions (rTHA) are available for THA. We aimed to estimate the cost-utility of cTHA compared to rTHA and mTHA in patients undergoing THA from the US healthcare system perspective. A Markov model was developed to compare costs and utilities of cTHA vs. mTHA, and cTHA vs. rTHA over a 1-year time horizon. Health states were defined based on the occurrence of readmissions with/without revisions due to fracture, dislocation, infection and hip pain. Utilities were presented in quality-adjusted life years (QALYs). Costs included length of stay, operative time and readmissions/revisions. The incremental cost-effectiveness ratio (ICER) was estimated as incremental cost per QALY change for each pairwise comparison. Inputs were drawn from published literature. cTHA was associated with a slight QALY gain of 0.001, and estimated savings of $1,595 and $949 per patient compared to rTHA and mTHA, respectively. Results indicated that cTHA was the ‘dominant’ strategy, i.e. reducing costs and slightly increasing QALYs, compared to both alternatives. Probabilistic sensitivity analysis indicated that cTHA was cost saving in 100% of the 1,000 simulations compared to both rTHA and mTHA. Using computer-assisted fluoroscopic navigation in THA showed cost savings and a slight improvement in quality of life compared to robotic-assisted and manual THA. Results suggest that computer-assisted fluoroscopic navigation is the preferred strategy for THA mainly due to downstream cost savings by reductions in OR time and readmissions/revisions rates.

ABSTRACT. The most common complication of femoral shaft treatment is a postoperatively detected torsional difference resulting into a revision surgery. This paper introduces an X-ray-based approach to determine femoral torsion intraoperatively. In this approach, the surgeon positioned the C-Arm to a specific standard projection at the proximal and the distal side of the femur. For the proximal side, the centerline of the shaft must pass through the center of the femoral head. For the distal side the condyles must to overlap completely. The rotation around the shaft axes is calculated by decomposing the orientations on both sides into the Euler angle separately. The difference between the roll angle at the proximal and the distal side of the femur defines the femoral torsion in our approach. The approach was evaluated using two left and one right human leg with acetabulum. The femoral torsion was determined five times on each specimen. The mean value and standard deviation were -22.2° ±0.8, -20.2° ±2.3, and 22.6° ±1.3 and the maximal deviation of 1.2°, 3.6°, and 2.2°. The results demonstrate that the introduced approach is capable to determine the femoral torsion on human femur. The shown performance indicates that torsion errors leading to revision surgeries (≥ 15°) could be recognized and avoided by this approach.

ABSTRACT. Background: Intraoperative imaging with mobile C-arms is standard in orthopeadic and trauma surgery, with increasing importance due to minimally invasive techniques. Usually, non-scrubbed operating room (OR) staff manage C-arm positioning. Motorized 3D C-arms, such as in hybrid ORs, allow remote control from the sterile field, reducing dependence on non-sterile personnel. This study compares a fully motorized mobile 3D C-arm (mC-arm) with a standard mobile 3D C-arm (sC-arm) in terms of intraoperative motion, radiation, and time parameters during lower extremity procedures in trauma surgery. Material & Methods: In a single-center observational study, 116 patients undergoing surgery on the tibia, ankle, or calcaneus were analyzed. Data on motion, radiation, and time were collected using log files, inertial measurement units, and dose reports. Key metrics included C- arm operating time (COT), C-arm operating ratio (COR), dose area product (DAP), and imaging operation (IO) duration. Results: No significant differences in COT, COR, or IO duration were found between the groups (all P>0.05). The DAP, number of images, and fluoroscopy time showed no significant differences between the imaging systems (all P > 0.05). All imaging in the mC-arm group was performed by the sterile surgical team without support from non-sterile personnel. Discussion: The fully motorized mobile 3D C-arm (mC-arm) demonstrates non-inferiority compared to the standard 3D C-arm (sC-arm) in intraoperative imaging for lower extremity trauma surgery, with no significant differences in time or imaging parameters. The mC-arm enables complete control from the sterile field, reducing reliance on non-sterile OR personnel. This highlights its potential to enhance workflow efficiency while maintaining imaging performance.

ABSTRACT. Metal artifacts significantly degrade the image quality of cone-beam computed tomography (CBCT), particularly in spine surgeries involving pedicle screws, complicating the assessment of implant positioning and surrounding anatomy. This study introduces a novel Metal Artifact Avoidance (MAA) workflow that leverages deep learning for trajectory optimization to reduce artifacts during CBCT acquisition. The automated approach incorporates real-time user verification, enabling tailored C-arm adjustments based on a predictive artifact model. The MAA workflow begins with scout views to detect metallic objects using a pretrained FasterRCNN model, followed by triangulation of their 3D positions. A physics-based artifact metric predicts the impact of various tilt angles, with the optimal trajectory suggested to the user through an intuitive visualization interface. The method is demonstrated on cadaveric data with pedicle screws in the lumbar spine, comparing standard and MAA-guided tilted scans. Results show that MAA-guided scans visibly reduced artifacts and enhanced visualization of critical anatomical structures, such as cortical surfaces around screws, compared to standard scans. The improvements achieved were consistent, even in cases where post-processing techniques like fsMAR failed to effectively mitigate artifacts. This study demonstrates that combining automated artifact prediction with user-verified trajectory adjustments provides a practical and reliable solution for artifact reduction. Future work will focus on validating the method on larger datasets and optimizing its integration into clinical workflows for broader adoption in spine surgery.

ABSTRACT. Developmental Dysplasia of the Hip (DDH) diagnosis relies heavily on traditional 2D ultrasound (US) imaging, which tends to be operator-dependent, highly variable and can often result in inconsistent diagnoses. DDH metrics automatically extracted using deep learning from 3D US scans have shown drastically improved diagnostic reliability in comparison to existing 2D US probes, but the high cost and limited availability of 3D US systems restrict their use in routine pediatric practice. We hypothesized that optically tracked 2D ultrasound (OTUS) systems, which have been previously demonstrated to have excellent spatial accuracy, could serve as a more accessible and reliable alternative. In this study, we collected OTUS scans from pediatric patients and evaluated repeatability through reconstructed volume similarity. Our findings show that OTUS achieves repeatability comparable to 3D US, highlighting its potential as a cost-effective and practical tool for improving the reliability of DDH screening.

ABSTRACT. Introduction In severe cases with lumbar lordosis loss > 25°, pedicle subtraction osteotomy is performed surgically to restore the sagittal malalignment. But, it has severe limitations. So, we are developing a new Y-shaped pelvic osteotomy that targets the pelvis rather than the spine. This work presents a fixation system tailored to the Y-shaped pelvic osteotomy. Methodology Criteria for the fixation system were set: Offer compressed fixation on the posterior side (closed wedge), maintain a 15° wedge opening on the anterior side and use conventional fixation components. This resulted in combining a lag-screw posteriorly and a patient-specific osteosynthesis plate anteriorly. To assess feasibility, a basic finite element analysis (FEA) was performed using loads derived from a static free body analysis, and a preliminary test was performed on a Sawbones model to asses surgical usability and osteotomy accuracy. Results The load input to the FEA was determined to be 397 N in x-direction, 832 N in z-direction, and 18 Nm moment. The FEA showed that the width of the T-shape of the plate crossing the osteotomy should be at least 15 mm to remain below allowable material stress level. The usability gave a score of 3.9 out of 7 for the fixation system. The Sawbone test showed obtained osteotomy angles of 16.6° and 19.5°, respectively compared to the set 15°. The bone contact area was 62% compared to the planned 61%. Conclusion The preliminary results indicate the feasibility of the Y-shaped pelvic osteotomy in combination with the new fixation system.

ABSTRACT. The deficiency of abductor mechanism of hip following total hip arthroplasty (THA) is rarely uncommon, but destructive and difficult to treat. The purpose of this study is to assess the contribution of the abductor mechanism of the hip and the gluteus maximus transfer (GMT) and tensor fasciae latae transfer (TFLT) on the stability of the total hip arthroplasty. The study subjects comprised 8 Thiel- embalmed cadaveric normal hip joints. We performed THA on 8 cadaveric specimen using CT based navigation. Posterolateral approach was used in this study. The deficiency model of abductor mechanism was created following THA. After that the GM and TFL muscle transfer was conducted. We measured the range of motion of the hip joint and the femoral head translation as indicators of soft tissue tension after each step by using the navigation system. There was a significant decrease in the range of motion for internal rotation after the gluteus maximus (GM) muscle transfer. The role of the GM muscle transfer in limiting internal rotation was confirmed. There was a significant decrease in the range of motion for external rotation after the TFL muscle transfer. The role of the TFL muscle transfer in limiting external rotation was confirmed. The femoral head translation significantly decreased after both muscle transfers, GM and TFL muscle transfer had the effect against hip traction of axis and lateral direction. The effect of the GM & TFL combined muscle transfer as a soft tissue reconstruction for THA with deficiency of abductor muscle was clarified. Each muscle transfer functions independently for internal rotation, external rotation. GMT effectively prevented excessive internal rotation, while TFLT prevented excessive external rotation.

ABSTRACT. Statistical shape models (“SSMs”) are powerful tools in computational orthopedics for analyzing anatomical variability and generating synthetic data for clinical applications. Existing models of the glenohumeral joint describe bones individually and are generally limited to non-pathologic datasets, failing to capture coupled shape variation in arthropathic anatomy. We introduce a novel combined two-body SSM for the scapula and proximal humerus, specifically addressing coupled variations in arthropathic glenohumeral anatomy. Using preoperative CT scans from a heterogeneous sample of 45 reverse total shoulder arthroplasty patients, the combined SSM integrates patient-specific anatomical relationships into predictive workflows.

Over 43% of the observed variation in scapula and proximal humeral anatomy was found to occur jointly, exposing a significant flaw in the independence assumption inherent to single-body models. The combined SSM outperforms single-body independent models in two key applications: predicting the shape of a missing body when presented with its counterpart, and generating varied, realistic, synthetic populations. Missing scapula predictions show a 10.09% error reduction when using the combined SSM, while de novo synthetic populations exhibit improved plausibility distributions, free from the generative biases caused by single-body SSMs.

Our findings emphasize the practical advantages of modeling anatomical dependencies for enhancing computational analyses and surgical planning. By addressing joint-level interactions, the combined SSM offers a more accurate and versatile framework for advancing patient-specific care in orthopedic surgery.

ABSTRACT. The hip joint center (HJC) does not only define the hip specific coordinate system in most definitions, but is also one of the most relevant functional parameters of the hip joint. Sphere fit of the femoral head is the most common method for HJC definition. In terms of preoperative planning for THA, many patients show deformities on the femoral heads and/or the acetabular region. Therefore, the approximation of the HJC via sphere fit has to be questioned. In our analysis, we studied different methods for automatic HJC definition and the influence of femoral head deformities on the location of the HJC for CT images of 201 THA patients. The different methods were ellipsoid fit on the femoral head, center of mass analysis of the femoral head, sphere fit on the acetabular region and geometric analysis based on ASIS location and pelvic width, height and depth. We compared the deviations between the sphere fit center and the different methods. We found best accordance with sphere fit for ellipsoid fit, followed by center of mass, acetabular sphere fit and geometric definition. The same tendency was found for the differences between sphere-like and deformed femoral heads, with deformed heads showing higher deviations for all methods. While no dynamic data was available, it has to be questioned, whether a sphere fit on the femoral head is the suitable for definition of HJC as center of rotation, especially for patients with deformed femoral heads. Further processing that takes different femur positions into account is recommended.

ABSTRACT. Stemless TSA requires sufficient bone density to ensure appropriate implant stability, both of which can be impacted by surgical precision. While bone density surrounding a stemless humeral implant and implant size are the strongest predictors of implant stability, this study shows that implant positioning also impacts bone density and, hence, stability. The increased precision offered by robotic surgery relative to conventional surgery is shown here to reduce the variability in bone density around the implant, and may therefore improve the primary stability of stemless TSA.

ABSTRACT. Unintended under- and overcorrections remain a significant challenge in medial open wedge High Tibial Osteotomy (owHTO). Achieving balanced load distribution is a central focus in HTO, yet most studies concentrate on the classical two-dimensional (2D) standing scenario rather than examining the complexities of three-dimensional (3D) load behavior during dynamic motions. This study aimed to investigate the biomechanical effects of key surgical parameters—wedge height, hinge axis, and osteotomy technique—on the position of the resultant force on the tibial plateau during knee flexion. A multibody simulation was conducted on 10 3D computer models of the tibia. The position of the center of pressure (CoP) on the tibial plateau was measured and compared across different surgical scenarios. Results indicate that increasing wedge height causes lateral CoP displacement, with the effect decreasing at higher flexion angles, while anteromedial axial rotation of the hinge axis led to posterior CoP shifts. A comparison of supratuberositary to infratuberositary osteotomies revealed a medial CoP displacement during late flexion (>5°).

ABSTRACT. Advancements in robotic-assisted TKA allow for controlled ligament loading before bony resections. The behavior of ligaments after preloading during TKA is highly variable and patient specific. Understanding these variations will help tailor loading protocols to optimize ligament tension. A retrospective review of 53 consecutive patients from a single surgeon was conducted to analyze ligament behavior after a cycling load using the BalanceBot system. Initial ligament laxity was recorded as the knee was ranged from flexion into extension, with 70-90N of force. Then the knee was loaded with 120 N for three flexion cycles. After the cycling ligament laxity was recorded under another 70-90N load. The ligament elongation was calculated from the initial to the final assessment. Ligament elongation was assessed at 10°, 45°, and 90°, with comparisons between lateral and medial performed using Wilcoxon rank sum and Levene’s tests. Sub-analyses were performed comparing males vs females. At 10° negligible elongation was observed medially or laterally (0.0±0.4 vs 0.1±0.5mm, p>0.05). At 45° both compartments showed similar elongation (0.2±0.6 vs 0.2±0.7mm, p>0.05). 90° saw the greatest elongation with lateral opening slightly more than medial (0.5±0.6 vs 0.4±0.4mm, p>0.05). At 45°, females exhibited significantly greater elongation than males, both laterally (0.3±0.5 vs. 0.1±0.6mm, p<0.05) and medially (0.4±0.3 vs. 0.0±1.0mm, p<0.05). The lateral side has a trend of opening more than the medial, and was more pronounced in females, particularly in midflexion. This highlights the importance of tailoring TKA procedures to individual patient characteristics.