← Back to home

Electromagnetic Navigation for Femoral Osteotomy Using High-Accuracy X-ray-to-CT Registration

Roman Flepp1,2*†, Arend Nieuwland1,3*†, Bastian Sigrist2, Philipp Fürnstahl2, Lilian Calvet2‡, Thomas Dreher1,3‡
1 Department of Pediatric Orthopedics and Traumatology, University Children's Hospital Zürich, Switzerland
2 Research in Orthopedic Computer Science (ROCS), University Hospital Balgrist, University of Zurich, Switzerland
3 Department of Orthopedic Surgery, University Hospital Balgrist, University of Zurich, Switzerland
* Corresponding authors: roman.flepp@kispi.uzh.ch, arend.nieuwland@balgrist.ch
These authors contributed equally.
These authors jointly supervised this work and share last authorship.

01. Overview

Accurate execution of preoperative plans in corrective femoral osteotomies remains a substantial challenge. Traditional free-hand execution depends heavily on surgeon experience and iterative fluoroscopy, exposing the patient and surgical team to significant radiation. While patient-specific 3D-printed guides (PSI) constrain tool placement to preoperative plans, they require extensive soft-tissue stripping, generate production lead times, and add to overall surgical costs.

To address these limitations, we present an integrated, electromagnetic tracking (EMT)-based navigation system. The workflow unifies three core components: preoperative CT-based planning, an automatic two-view C-arm calibration and X-ray-to-CT registration pipeline, and real-time intraoperative tracking of surgical tools. This system maintains a minimal surgical footprint and provides real-time guidance for sawblade positioning and fragment correction without line-of-sight constraints.

  • Integrated EMT navigation pipeline combining CT-based preoperative planning, one-time C-arm calibration, and X-ray-to-CT registration from only two fluoroscopic images.
  • 52% lower mean angular error than free-hand technique (3.05° ± 0.75° vs. 6.32° ± 2.36°, p = 0.031), with zero EMT trials exceeding the 5° clinical threshold.
  • Statistical equivalence to patient-specific instrumentation (PSI) in angular (p ≤ 0.02) and translational (p = 0.048) accuracy, without PSI's added soft-tissue exposure or fabrication lead time.
  • U-Net-based automatic C-arm calibration from fluoroscopic bead detection, replacing manual calibration steps.

02. Method & Pipeline

Method overview pipeline
Figure 1: System dataflow overview. The workflow starts from preoperative CT planning (left), proceeds to automatic intraoperative calibration and registration using two fluoroscopic views (middle), and finishes with dynamic real-time sawblade and fragment tracking displayed on the custom GUI (right).

The registration protocol operates in two phases:

Initial Registration: Establishes a spatial link between the preoperative CT bone model and the EMT coordinate system using a custom calibration sleeve mounted with EMT sensors. A U-Net-based model automatically identifies calibration bead locations in the C-arm images to solve C-arm intrinsic/extrinsic parameters and compute the 3D X-ray-to-CT transform.

Real-Time Navigation: Tracks the relative pose of distal fragments with respect to the proximal femur using miniature sensors anchored to 3mm Kirschner wires (K-wires). Dynamic updates enable the visual GUI to continuously decompose alignment errors into extension, derotation, and varisation.

03. Navigation Animation

Demonstration Video: Real-time rendering of osteotomy cut navigation and bone alignment tracking. The GUI maps tool tips, sawblade orientations, and computed corrective trajectories, allowing surgeons to dynamically visualize and verify the bone adjustment relative to the target plan.

04. Key Feasibility Results

The proposed system was validated through a feasibility study utilizing 18 synthetic femora replicas and compared against conventional free-hand techniques and patient-specific instrumentation (PSI) guides.

Method Mean Angular Error (°) Mean Translational Error (mm) Outliers (>5° Clin. Threshold)
Free-hand 6.32 ± 2.36 5.56 ± 2.25 4 / 6 trials
PSI (Guides) 3.44 ± 1.16 3.36 ± 1.18 0 / 6 trials
EMT Navigation (Ours) 3.05 ± 0.75 3.59 ± 1.44 0 / 6 trials

Our navigation system showed a 52% reduction in mean angular error compared to free-hand techniques. Crucially, the system demonstrated statistical equivalence to PSI guides in both rotation and translation accuracy, while requiring significantly less soft-tissue exposure and only two fluoroscopic views for initialization (compared to clinical free-hand benchmarks which average over 34 images).

05. Citation

A machine-readable Markdown version of this page (abstract, key contributions, results table, term definitions, BibTeX) is available at papers/electromagnetic-navigation-femoral-osteotomy.md.

@article{flepp2026emtosteotomy,
  title   = {Electromagnetic Navigation for Femoral Osteotomy Using High-Accuracy X-ray-to-CT Registration},
  author  = {Flepp, Roman and Nieuwland, Arend and Sigrist, Bastian and Fürnstahl, Philipp and Calvet, Lilian and Dreher, Thomas},
  journal = {arXiv preprint arXiv:2606.03893},
  year    = {2026},
  url     = {https://arxiv.org/abs/2606.03893}
}