Digital Subtraction Angiography Based On Image Features Registration

Digital Subtraction Angiography is originated in 1980's as a medical imaging technology. It is an important part in Digital Radiography. But for the fluctuations in the power of the X-ray source, the noise in the image intensifier, the patient motion and peristaltic motion of organ, artifacts are caused in DSA. In this paper, it researches motion correction techniques to reduce artifacts in the result image. Following is the major job list.1) In most image registration algorithms, regular grid point is regarded as control point (CP), which doesn't consider the image feature at all. Reversely, in the paper, it fully takes advantage of the image feature such as corner and edge, and chooses CPs on those positions. With the 3D space-time algorithm, the CPs of the contrast vessel can be reduced or eliminated. The extracted CPs will make window registration more accurate and efficient.2) Because traditional similarity measure Energy of the Histogram of Differences (EHD) has to compute square, it is computationally expensive. In the paper, it takes improved PAIRS for less computation time instead of EHD. and chooses an optimization strategy for computation in measure of match.3) Even subpixel wrong-registration can cause significant artifacts in the subtraction images. It obviously must provide means to reach subpixel precision in the displacement computations. This requires an interpolation method.4) After geometrically corrected by affine transformation, mask image performs the global gray correction to further eliminate fluctuations in the intensity of X-ray and the nonuniform diffusion of the contrast medium into the capillaries.