Segment self-guide reconstruction algorithm based on object-oriented quantization

Aiming at the problem of inaccurate imaging model of three-dimensional (3D) reconstruction of rotational DSA (digital subtraction angiography) images, firstly a nonlinear model based on object-oriented quantization is introduced. The model quantizes the projective pixel of 3D vessel slice as the vessel number that the X-ray goes through. Then, under the constraint of limited views and sparse projections, a slice reconstruction algorithm named segment self-guide reconstruction (SSGR) is developed. It converts the slice reconstruction of N+1 level nonlinear quantized DSA image to the reconstruction of N vessel cross-sections. The SSGR is especially suitable for solving the problem of sparse projections and limited-views. Finally, the simulated results have proved the feasibility of the model and the validity of the algorithm.