Removal of bone information in dual energy CT

Computer tomography angiography (CTA) allows to assess the degree of stenosis in vascular systems, but requires the separation of vessels against bone and calcifications. In this paper, a new technique of converting dual energy data into a virtual native and contrast agent scan pair allows to use a graph based method of bone removal for CTA. For evaluation purposes, patients with a high degree of calcification, i.e. complicated cases, are considered. Our automatic bone-removal technique requires 28-30 minutes on a Intel Core 2 Quad 2.66 GHz with 4 GByte RAM for a 1.2 GByte data set and is applied in the clinic for evaluation purposes. For 10 evaluated patients, sensitivity, specificity, and accuracy are 90%. A further improvement of the differentiation quality is limited by the noise of the CT images.