Session S94.2

2D-3D Registration of Cardiac Images Using Catheter Constraints

MVN Truong*, A Aslam, CA Rinaldi, R Rezavi,
G Penney, K Rhode

King's College London
London, UK

Cardiac catheterisation procedures are routinely guided by X-ray fluoroscopy, but suffer from poor soft tissue contrast and a lack of depth information. These procedures can employ magnetic resonance (MR) or computerised tomography (CT) imaging pre-operatively for treatment planning due to their excellent soft-tissue contrast and 3D imaging capabilities.
The aim of our work was to combine the benefits of both X-ray fluoroscopy and 3D imaging modalities to guide cardiac catheterisation procedures. A 2D-3D feature-based image registration method was developed to overlay pre-operative MR/CT images onto the intra-operative X-ray fluoroscopy to act as a roadmap.
Our method was applied to electrophysiology procedures, which have a high failure rate and therefore would benefit most from guidance. During these procedures, catheters are routinely placed in the heart and great vessels, in particular the coronary sinus and the aorta, where they remain throughout the catheterisation procedure. These vessels were segmented from the pre-operative 3D anatomical scans, MR data for our study, and the intra-operative 3D catheter positions were reconstructed from a pair of sequential biplane X-ray images. The MR and X-ray images were gated at end-diastole and end-expiration.
To achieve registration, an automatic global-fit algorithm was developed which used the constraint that the catheters must lie close, i.e. minimise root mean square distance error, to the medial line of their respective segmented vessels.
Data were processed for one clinical case where an accurate registration was validated using visual inspection by an expert. Further validation on more clinical cases is currently being carried out.

(Abstract Control Number: 220)