3D reconstruction of coronary veins from a single X-ray fluoroscopic image and pre-operative MR

Cardiac resynchronization therapy (CRT) is an effective treatment for patients with congestive heart failure and ventricular dyssynchrony. Despite the overall efficacy of CRT, approximately 30% of patients receiving CRT do not improve. One of the main technical problems related to the CRT procedure is inadequate visualisation in X-ray fluoroscopy of the venous anatomy in relation to accurate cardiac chamber visualisation. This paper proposes a novel approach for 3D reconstruction of coronary veins from a single contrast enhanced intra-operative fluoroscopy image. For this application, the method uses back-projection geometry and a Euclidean distance/angle-based cost function. The algorithm is validated on a phantom and five patient datasets, comprising six view-angle orientations for the phantom dataset and two view-angle orientations for each of the patient datasets. Median(inter-quartile range) 3D-reconstruction accuracies of 1.41(0.55–3.00) mm and 3.28(2.10–4.89) mm were established for the phantom and patient data, respectively. The technique can facilitate careful advancement of the cannulating guide over a guidewire or a diagnostic catheter positioned in the coronary sinus, and consequently, improve the chances of response to CRT.