Atrial fibrillation is the most common sustained heart arrhythmia and a leading cause of stroke. The treatment by catheter ablation, performed using fluoroscopic image guidance, has become the treatment of choice. Such procedures are performed in electrophysiology labs equipped with modern C-arm X-ray systems. Fluoroscopic overlay images rendered from pre-operative volumetric data, e.g., CT, MRI or C-Arm CT, provide additional guidance for physicians during these catheter ablation. As these overlay images are compromised by cardiac and respiratory motion, we propose a method to provide motion compensation by tracking a commonly used mapping catheter.

A catheter model is generated by manual initialization. The catheter is segmented by a boosted classifier cascade. Afterwards, a model-based registration is applied to perform motion compensation. We achieve a frame rate that is higher than the clinical demands and our accuracy is as low as the thickness of the catheter.