Technical note: unsupervised C-arm pose tracking with radiographic fiducial

Area of Science:

• Medical Imaging
• Radiological Reconstruction
• Surgical Navigation

Background:

• Accurate C-arm fluoroscopy imaging is crucial for 3D reconstruction in procedures like prostate brachytherapy.
• Current methods often require manual segmentation of radiographic fiducials, which is time-consuming and prone to error.

Purpose of the Study:

• To develop an automated and unsupervised method for C-arm pose tracking.
• To eliminate the need for prior segmentation of radiographic fiducials in C-arm images.

Main Methods:

• Computed C-arm poses relative to a stationary radiographic fiducial of known geometry.
• Registered filtered 2D fluoroscopy images to their 3D models using image intensity without segmentation.
• Investigated a three-step cascade filter and a line enhancement filter for image enhancement.

Main Results:

• Evaluated 2775 cases using 111 clinical C-arm images with perturbations.
• Achieved average rotation errors of 0.62 degrees (three-step filter) and 0.67 degrees (line enhancement filter).
• Achieved average translation errors of 0.72 mm (three-step filter) and 0.87 mm (line enhancement filter).

Conclusions:

• The developed C-arm pose tracking method is accurate and robust for human patient data.
• This automated approach facilitates subsequent 3D implant reconstruction.
• The method shows promise for improving efficiency and accuracy in image-guided interventions.