Image-based RSA: Roentgen stereophotogrammetric analysis based on 2D-3D image registration

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Orthopedics

Background:

• Conventional Roentgen stereophotogrammetric analysis (RSA) relies on radiopaque bone markers for 3D motion tracking.
• Limitations exist in marker placement and potential for marker migration, impacting accuracy and applicability.
• There is a need for markerless techniques to enhance RSA's utility in clinical and research settings.

Purpose of the Study:

• To introduce and validate Image-based Roentgen stereophotogrammetric analysis (IBRSA) as a novel markerless 3D motion analysis method.
• To assess the accuracy and precision of IBRSA in comparison to conventional RSA.
• To explore the potential applications of IBRSA in research and clinical practice.

Main Methods:

• IBRSA integrates 2D-3D image registration using 3D CT data and digitally reconstructed radiographs (DRRs).
• The 3D pose of the CT data is iteratively adjusted to match 2D RSA images via an image matching metric.
• Validation was performed on cadaveric and sawbone scapula radiographs using RSA and a micromanipulator as gold standards.

Main Results:

• IBRSA achieved good matching results in validation experiments.
• Accuracy: |mu|<0.083 mm (translations), |mu|<0.023 degrees (rotations).
• Precision (sigma): 0.090, 0.077, 0.220 mm (translations); 0.155, 0.243, 0.074 degrees (rotations).
• In vitro IBRSA accuracy/precision is lower than in vitro RSA but higher than in vivo RSA.

Conclusions:

• IBRSA offers a viable markerless alternative for 3D motion analysis, leveraging 3D CT data.
• The method demonstrates promising accuracy and precision, suitable for specific research applications.
• Eliminating the need for radiopaque markers opens new avenues for dynamic analyses and computer navigation in orthopedics.