Vincent M. A. Stirler

Three-dimensional (3D) printed, patient-specific splints may offer an alternative for conventional immobilization methods by enhancing comfort, aeration, and hygiene. However, their broader clinical use remains limited by the complexity of current design workflows. Most existing workflows depend on 3D scans processed with commercial software and on manual or semi-automated CAD procedures that are difficult to reproduce, time-intensive, and often require pot-fit adjustments that are challenging to implement. A scalable modeling framework capable of capturing the substantial anatomical variability of the human forearm is needed to support more automated design pipelines. Morphable models address these limitations by establishing dense point-to-point correspondence across anatomical meshes, enabling a reference shape with fixed topology to deform through a set of parameters to approximate individual anatomies accurately. This study developed and validated a morphable forearm model using 200 structured light scans from 100 healthy adults (41 ± 19 years). Mesh preprocessing, anatomical alignment, and iterative registration were performed using a standardized framework, after which a statistical shape model was constructed and evaluated by comparing deformed mean shapes to original geometries using vertex-to-surface distance metrics. The model achieved a mean reconstruction error of 0.05 ± 0.11 mm, with the largest deviations near mesh boundaries at the hand and upper arm. These results demonstrate that the morphable model accurately represents forearm shape variability and provides a reproducible foundation for semi-automated, patient-specific splint design, thereby supporting the more efficient and clinically scalable implementation of 3D-printed immobilization solutions.

Triage of patients in the emergency department is a complex logistical process, and diagnostic errors are more likely to occur there than at any other hospital department. Automated assessment of injury severity, need for intervention and likelihood of complications could improve patient flow and reduce unnecessary invasive interventions or intensive care admissions. This proof-of-principle study aimed to evaluate the performance of a deep learning model in the detection of liver injury as a first step in the development of an automated assessment system. The model was based on an nnU-net architecture, modified to classify computed tomography images as injured or uninjured and segment the identified injured cases. The dataset was comprised of patients from the emergency department of the Radboud University Medical Centre, a level 1 trauma centre and military affiliation hospital. The training dataset contained 88 patients, of whom 44 patients with trauma-related liver injury and 44 without liver injury. The model was evaluated on a test set comprised of 10 patients with injured and 10 patients with uninjured livers. The proposed model achieved a classification accuracy of 0.85, sensitivity of 0.9 and specificity of 0.8. The segmentation of the identified injured cases achieved Dice/recall/precision scores of 0.948/0.945/0.948 for uninjured and 0.818/0.786/0.925 for injured parenchyma. The study showed that the model can classify patients and segment the liver parenchyma and injured regions and proves a good first step in the development of an algorithm to assist clinicians in the detection and assessment of liver injuries after high-energetic trauma.

Smits I, et al. Clin Orthop Surg. 2025 Oct;17(5):747-753. https://doi.org/10.4055/cios24418

Background: Adequate pain relief, early restoration of breathing, and rapid mobilization pose a clinical challenge in patients with blunt chest trauma. Virtual reality (VR) has the potential to achieve these 3 interrelated treatment objectives with enhanced self-efficacy and autonomy of patients and limited support by clinicians. Objective: This study aimed to assess the effectivity of breathing and physical exercises using VR on the pulmonary recovery of patients with blunt chest trauma at the ward. Methods: A pilot randomized controlled trial was performed. The control group received usual physiotherapy consisting of protocolized breathing exercises (8 times daily for 10 minutes) and physical exercises (2 times daily for 10 minutes). The VR group was instructed to perform these exercises using VR. The primary outcome was vital lung capacity at day 5 or earlier at discharge. Secondary outcomes were patient mobility (time standing, lying, and sitting), clinical outcomes (length of hospital stay, pulmonary complications, transfer to intensive care unit, and readmission within 30 days), pain, activities of daily living, patient-reported outcome measures (satisfaction and quality of recovery). Patient experiences and barriers and facilitators toward implementation were assessed through interviews. Results: The study was prematurely ended due to enrollment failure combined with poor protocol adherence to exercises in both groups. A total of 27 patients were included, of which 19 patients completed 3 or more days. Vital lung capacity at 5 days (or last measurement) was equal between groups with 1830 (SD 591) mL and 1857 (SD 435) mL in the control and VR groups, respectively. No marked differences were observed in secondary outcomes. Patient interviews showed positive attitudes toward the use of VR, describing that visualization of the exercises helped patients to perform the exercises correctly and to continue the exercises for a longer duration. Also, patients experienced the immersiveness of VR as an analgesic. However, patients did not experience added value over usual care and reported that better integration in treatment and the hectic hospital environment could improve the use of the VR exercises. Conclusions: The suitability of patients to use virtual reality therapy (VRx) in a hospital (trauma) ward setting is lower than generally expected. Effective application of VRx requires professional guidance and needs thorough alignment with clinical practice. For future research, we recommend to chart adherence to study protocol before designing a VR clinical trial. Patient-reported experiences need to be prioritized in evaluating VR acceptance, usability, and effectiveness. In line, we recommend performing a systematic analysis (eg, using the technology acceptance model) on the acceptance before pilot or main effectiveness studies. Finally, the eligibility of patients and exclusion of patients due to the inability to use VRx should be routinely reported. Trial Registration: ClinicalTrials.gov NCT05194176; https://tinyurl.com/2bzh4tzx

The conventional treatment for distal radius fractures typically involves immobilization of the injured extremity using a conventional forearm cast. These casts do cause all sorts of discomfort during wear and impose life-style restrictions on the wearer. Personalized 3D printed splints, designed using three-dimensional (3D) imaging systems, might overcome these problems. To obtain a patient specific splint, commercially available 3D camera systems are utilized to capture patient extremities, generating 3D models for splint design. This study investigates the feasibility of utilizing a new camera system (SPENTYS) to capture 3D surface scans of the forearm for the design of 3D printed splints. In a prospective observational cohort study involving 17 healthy participants, we conducted repeated 3D imaging using both the new (SPENTYS) and a reference system (3dMD) to assess intersystem accuracy and repeatability. The intersystem accuracy of the SPENTYS system was determined by comparison of the 3D surface scans with the reference system (3dMD). Comparison of consecutive images acquired per device determined the repeatability. Feasibility was measured with system usability score questionnaires distributed among professionals. The mean absolute difference between the two systems was 0.44 mm (SD:0.25). The mean absolute difference of the repeatability of the reference -and the SPENTYS system was respectively 0.40 mm (SD: 0.30) and 0.53 mm (SD: 0.25). Both repeatability and intersystem differences were within the self-reported 1 mm. The workflow was considered easy and effective, emphasizing the potential of this approach within a workflow to obtain patient specific splint.