Reconstruction of blood vessels from X-ray substraction projections

The author shows that it is possible to apply a priori knowledge in X-ray image processing to isolate particular anatomy, provided that anatomy is characterized by a unique enough signature. In the case of 3-D vasculature, two pieces of a priori knowledge can be brought to bear. First, image subtraction can be used to discriminate between opacified vessels and nonopacified tissues. Second, vessels can be better isolated in 3-D if it is assumed that vessels are most likely surrounded by a void. In the case of focal plane digital subtraction angiography, significant improvement in image quality is achieved by using this information judiciously.<>