Improved Reconstruction of Highly Under-Sampled MR Angiography Images using Modified Nonlocal Means

Yanqiu Feng¹, Yingjie Mei¹, Cong Wang¹, Wufan Chen¹

In this paper, a novel compromise between spatial and temporal resolution is proposed based a modified nonlocal-means (NLM) filtering. Temp images are reconstructed using gridding algorithm from data being combined along temporal direction with the rectangle window function. Narrow window will produce HSR image and wide window will produce HTR image. Then, high spatio-temporal resolution is accomplished through a novel combining of the HTR image and the HTR image based on the modified NLM method. Preliminary experiments of simulated data demonstrate that the proposed method is of great potential for the reconstruction of highly under-sampled MR Angiography.