基于造影图像的能量场反投影合成冠脉三维重建

Abstract

3-D reconstruction of coronary artery has important clinical value for diagnosis and interventional treatment of cardiovascular diseases. In this paper, a novel energy back-projection composition model (EBPCM) is developed for reconstructing coronary arteries from two angiographic images. First, the imaging principle of X-ray and open parameter curve model are studied in detail. Then, the external forces of the parameter curves in the two angiographic images are back-projected into 3-D space and composited. Under joint effects of external driving forces and internal smoothing forces, the parameter curve is evolving continuously in 3-D space towards the true vascular structure with minimum composited energy. In order to improve the reconstruction accuracy, curve alignment method is employed to find the point-to-point matching relationship in the two angiographic images. Based on the fine points matching, bundle adjustment is utilized to perform non-linear optimization of projection parameters and the vascular structures simultaneously. The proposed method is validated on both phantom data and real angiographic images and the reconstruction results are compared on images with different imaging angles. Experimental results show that the developed method is very effective and robust, which can obtain accurate reconstruction of coronary arteries with space RMS error less than 0.80 mm.