C-arm Pose Estimation Using a Set of Coplanar Ellipses in Correspondence

Area of Science:

• Medical Imaging
• Surgical Navigation
• Computer Vision

Background:

• C-arms are vital for surgical procedures, requiring accurate pose estimation for quantitative feedback.
• Non-encoded C-arms, common in operating rooms, lack precise position data.
• Accurate C-arm tracking is essential for enhanced surgical guidance and safety.

Purpose of the Study:

• To develop an accurate and fast pose estimation method for non-encoded C-arms.
• To improve the robustness of C-arm tracking against noise in fluoroscopy imaging.
• To provide reliable quantitative feedback during surgical interventions using C-arm imaging.

Main Methods:

• Utilized an image-based fiducial comprising coplanar ellipses for C-arm tracking.
• Adapted a planar homography method with three key modifications: weighted linear system, orthonormalized rotation vectors, and constrained optimization.
• Simulated realistic X-ray imaging and performed sensitivity analyses for segmentation and calibration errors.

Main Results:

• Achieved mean tracking accuracy within 0.5 degrees and 0.9 mm.
• Demonstrated robustness to segmentation error variance up to 2 pixels squared.
• Showed resilience to calibration errors up to 1 cm, indicating practical applicability.

Conclusions:

• The proposed modified homography method enhances C-arm pose estimation accuracy and robustness.
• The technique is suitable for non-encoded C-arms, offering valuable quantitative feedback in surgery.
• The fiducial-based tracking system shows promise for improving surgical navigation and outcomes.