Simultaneous 3D-2D image registration and C-arm calibration: Application to endovascular image-guided interventions

Area of Science:

• Medical Imaging
• Image Registration
• Interventional Radiology

Background:

• 3D-2D image registration is vital for fusing pre-interventional 3D and intra-interventional 2D images.
• Accurate registration is essential for image guidance in procedures like intracranial endovascular image-guided interventions (EIGIs).
• Registration accuracy is influenced by the method, initial pose uncertainty, C-arm geometry, and the number of 2D images used.

Purpose of the Study:

• To evaluate the accuracy and robustness of 3D-2D image registration methods for EIGIs.
• To validate registration methods against a gold standard using clinical datasets.
• To assess the impact of initial registration stages and C-arm calibration on overall accuracy.

Main Methods:

• The study employed a two-stage registration process: initial (machine-based or template matching) and final.
• Four state-of-the-art intrinsic image-based 3D-2D registration methods were evaluated in the final stage.
• Objective validation was performed on 15 patients undergoing cerebral EIGI, with gold standard registrations established via fiducial markers.

Main Results:

• Success rates for 3D-2D registration improved from 36% (initial machine-based) to 73% (template matching) and 93% (final registration with C-arm calibration).
• The best registration accuracy achieved was 0.59 mm after final registration.
• Simultaneous C-arm calibration with 3D-2D registration consistently improved success rates, especially when compensating for in-plane translation errors.

Conclusions:

• Tested methods offer high potential for automation due to simultaneous C-arm calibration and 3D-2D registration using anatomical information.
• These methods can be readily integrated into current interventional workflows.
• Comprehensive validation under realistic clinical conditions was a key contribution.