
Background Population-wide osteoporosis screening is hindered by limited access to Dual-energy X-ray Absorptiometry (DXA). Opportunistic screening using plain radiographs is a promising alternative, but safe clinical adoption requires models that can reliably quantify their uncertainty. We evaluated a deep learning model that estimates femoral-neck bone mineral density (BMD) from knee radiographs and employs conformal prediction to provide calibrated, patient-specific uncertainty intervals. Methods We utilized bilateral knee radiographs with paired femoral-neck DXA scans (within 180 days) from the Osteoarthritis Initiative. The dataset was split at the participant level into training (70%), validation (10%), test (10%), and calibration (10%) sets. An EfficientNet-V2-M model was trained to predict BMD. We implemented Split Conformal Prediction with two test-time augmentation (TTA) strategies: (1) averaging augmented predictions before conformalization (Standard TTA), and (2) treating each augmented view as a separate sample (Multisample TTA). Performance was evaluated using Pearson correlation ( r ), mean absolute error (MAE), and empirical coverage of 90%, 95%, and 99% prediction intervals. Results The model achieved a Pearson correlation of r = 0.6800 and MAE of 0.1107 g/cm 2 with standard TTA. Conformal prediction successfully produced valid uncertainty intervals, with empirical coverage matching nominal levels (e.g., ∼95% coverage for 95% intervals). Multisample TTA yielded slightly narrower 95% intervals for the bilateral model (0.2776 vs. 0.2792 g/cm 2 ) without sacrificing coverage. Crucially, interval widths correlated with prediction errors, indicating the model correctly signaled higher uncertainty for difficult cases. Conclusion Deep learning can extract BMD signals from knee radiographs, and conformal prediction provides a rigorous “safety wrapper” by quantifying uncertainty. This approach offers a foundational step towards trustworthy opportunistic osteoporosis screening in orthopedic settings.
Background Management of concomitant scapholunate (SL) injuries with distal radius fractures (DRFs) remains controversial. This study evaluates whether open reduction and internal fixation (ORIF) of DRF alone restores the SL relationship in moderate SL diastasis. Methods This retrospective multi-center study included DRF patients with moderate SL diastasis (> 2.50 mm). Scapholunate distance (SLD) and angle were measured post-injury, post-operatively, and at final follow-up. Data were analyzed using a linear mixed model. Results ORIF of the DRF alone significantly reduced mean SLD from 2.96 mm (post-injury) to 1.94 mm (post-operatively). The SL angle showed no significant variation, with no progression toward dorsal intercalated segmental instability. Conclusion Findings suggest moderate SL diastasis in DRF often representing carpal extrinsic ligament malalignment rather than SL interosseous ligamentous failure. ORIF is effective in restoring the SL relationship by anatomical bony reduction. Additional immediate ligamentous repair may not provide clinical advantages over primary osseous fixation alone.
Background The use of three-dimensional (3D) navigation percutaneous screw fixation has been reported in non-displaced or minimally displaced pelvi-acetabular fractures. For carefully selected displaced acetabular fractures, 3D navigation technique can be applied, which traditionally required open approach. Methods From 2018 to 2024, 29 patients underwent 3D-guided navigation percutaneous screw fixation. Various reduction techniques were used. Functional and radiological outcomes were assessed. Results Twenty-four patients were followed up after definitive fixation with average duration of 33 months. All patients showed fracture union with average union time of 7.3 months. The mean Merle d’Aubigne and Harris Hip scores were 14.3 and 86.2 points, respectively. Younger age, lower injury severity score and earlier definitive operation correlated with improved functional outcome. Percutaneous fixation demonstrates advantages in terms of operative duration, blood loss, length of stay. Conclusion This surgical technique safely and effectively manages pelvi-acetabular fractures, resulting in pain relief, early mobilization and favourable patient outcomes.
Background In-person therapy (IPT) after orthopedic surgery can be costly, time-consuming, and difficult to access, particularly for patients affected by social determinants of health. Free web-based therapy (WBT) programs may offer a more accessible alternative. This study evaluated patient interest, barriers, and preferences for WBT following shoulder or wrist surgery. Methods A cross-sectional survey was administered to 50 postoperative patients at a level 1 trauma center during 2- or 6-week follow-up visits, assessing willingness to use WBT, barriers to IPT and WBT, and desired program features. Results Participants were 18–70 years old; 60% underwent wrist and 40% shoulder surgery. Fifty-eight percent had not attended IPT, most commonly due to time, cost, and distance. Interest in WBT was high (78%), particularly for free and customized programs. Reported WBT barriers included limited personal interaction, computer access, and customization. Conclusions Patients demonstrated a strong interest in WBT, especially when free and personalized. Virtual platforms may help address common barriers to postoperative therapy, though concerns regarding personalization, interaction, and technology access remain important considerations.
Introduction : Total hip arthroplasty (THA) is a common orthopaedic procedure that significantly alleviates pain and enhances the quality of life for patients with hip osteoarthritis. Precise implant placement is critical for successful outcomes. Robotic-assisted THA has been developed to improve surgical accuracy and predictability, offering three-dimensional computed tomography-based planning and haptic guidance during key steps of surgery such as reaming and cup impaction. This study aims to compare (i) the accuracy of implant positioning between robotic-assisted and conventional THA, and (ii) other intra-operative and post-operative outcomes, including blood loss, operative time, length of stay, and associated complications. Materials and Methods : In total, 97 THA cases performed between November 2017 and October 2023, with a minimum follow-up of one year, were retrospectively analyzed. Fifty patients underwent robotic-assisted THA, while 47 received conventional THA. Demographic data, radiographic outcomes, surgical time, blood loss, length of stay, and surgical complications were evaluated. Intra- and Inter-observer reliability of radiographic measurements was confirmed. Results : Robotic-assisted THA demonstrated significantly greater accuracy and consistency in achieving targeted acetabular inclination and anteversion (p < 0.001). More cases in this group fell within the Lewinnek (<0.001) and Callanan safe zones (<0.001). Improvements in femoral offset ( p < 0.001) and maintenance of global offset ( p = 0.264) were also observed. More cases achieved balanced limb length discrepancy ( p = 0.047). No significant differences were found in operative time ( p = 0.196), blood loss ( p = 0.775), or length of stay ( p = 0.236). Similar rates of surgical complication were observed in both groups. Discussion and conclusion : Robotic-assisted THA provides more precise and consistent implant positioning without increasing operative time, blood loss, length of hospital stay, or associated complications.