Registration for DSA Image Using Triangle Grid and Spatial Transformation Based on Stretching
Ying Chu, Nan Bai, Zhen Ping Ji, Siping Chen, Xuanqin Mou · 2006
Triangle grid and a spatial transformation method based on stretching were used to implement the registration for digital subtraction angiography (DSA) image to reduce the artifact caused by motion displacement. Block matching was used to calculate the motion displacement of each pixel in given mask and live images. Computational overhead was reduced by calculating only reference points for control points by block matching. Control points were selected in mask image according to an edge-detection scheme. The remaining pixels in calibrated image were mapped by spatial transformation and interpolation of the calculated reference points. The presented spatial transformation method based on stretching solves unexpected ill-behaved solutions common to traditional linear space transformations. Mathematical analysis of this stretching-based space transformation shows it like a linear transformation. Investigation confirms this spatial transformation method to be simpler and more efficient than traditional linear transformation methods. Motion displacement was better compensated and the quality of DSA images after registration was significantly improved