Ultrasound-Guided Percutaneous Scaphoid Pinning: Operator Variability and Comparison with Traditional Fluoroscopic Procedure

Maarten Beek, Purang Abolmaesumi, Suriya Luenam, Richard W. Sellens, David R. PichoraMICCAI 2006 · DOI

Introduction

The scaphoid bone, a small boat-shaped structure in the wrist, is the site of approximately 90% of all wrist fractures. Due to its limited blood supply, these fractures are prone to non-union, often resulting in chronic pain and arthritis. Surgical intervention typically involves driving a screw through the "waist" of the bone. While percutaneous (through the skin) pinning is less invasive than open surgery, it is traditionally performed "blindly" using 2D fluoroscopy (X-ray). This standard approach suffers from radiation exposure, the difficulty of interpreting 3D trajectories from 2D snapshots, and anatomical crowding from adjacent wrist bones.

Proposed 3D Ultrasound-Guided Workflow

To address these limitations, researchers presented a computer-assisted surgical (CAS) workflow at MICCAI 2006 that replaces fluoroscopy with 3D ultrasound-guided navigation. The process begins with pre-operative CT modeling to create a high-fidelity 3D surface of the scaphoid. During surgery, a 3D ultrasound transducer captures the bone's surface, which is then registered to the CT model using the Iterative Closest Point (ICP) algorithm.

An optical tracking system monitors the wrist and the surgical drill in real-time, allowing the surgeon to follow a virtual path on a monitor. This interface provides a direct 3D "GPS" for the procedure, significantly reducing the cognitive load required to mentally reconstruct bone anatomy from 2D images.

The graphical interface provides real-time 3D navigation, showing the drill's position relative to the planned trajectory on the scaphoid model.

Experimental Validation and Results

The system was evaluated using 16 human scaphoid phantoms (Sawbones) equipped with alumina beads for ground-truth measurement. The study compared the performance of two operators using the ultrasound system and benchmarked one operator against traditional fluoroscopy.

The results demonstrated that ultrasound guidance offers superior accuracy and consistency. For the entry point, ultrasound achieved a mean error of 1.89 mm, compared to 2.63 mm for fluoroscopy. The improvement was even more pronounced at the exit point, where ultrasound recorded a 3.04 mm error versus 5.73 mm for the traditional method.

Comparison of entry and exit point errors (mm) between ultrasound-guided and traditional fluoroscopic methods, showing reduced mean error and variability for the ultrasound approach.

Beyond raw accuracy, the ultrasound method showed tighter groupings in results, which is critical in surgery to avoid "outliers" where a screw might exit the bone prematurely. Furthermore, the system eliminated the need for intra-operative radiation (averaging 7.6 X-ray images per traditional procedure) and maintained a clinical workflow efficiency with registration times under three minutes.

Discussion and Future Directions

While promising, the study identified several areas for improvement. The ICP algorithm currently requires a manual initial alignment to avoid local minima. Future iterations may incorporate landmark-based registration to automate this step. Additionally, while the phantom study proved the concept, further research is needed to evaluate performance in the presence of soft tissue attenuation and potential bone mobility in vivo. Despite these limitations, the transition to 3D ultrasound navigation represents a significant step toward safer, more consistent orthopedic interventions.

Summary updated

BibTeX
@inproceedings{beek2006ultrasound,
  author = {Maarten Beek and Purang Abolmaesumi and Suriya Luenam and Richard W. Sellens and David R. Pichora},
  title = {{Ultrasound-Guided Percutaneous Scaphoid Pinning: Operator Variability and Comparison with Traditional Fluoroscopic Procedure}},
  booktitle = {International Conference on Medical Image Computing and Computer-Assisted Intervention},
  year = {2006},
  doi = {10.1007/11866763\_66},
}
Figure