Abstract #2872

Data Driven Reconstruction of Inconsistent K-Space Data

Kevin Michael Johnson¹, Walter F. Block^1,2, Scott B. Reeder^1,3, Alexey Samsonov^1,3

¹Medical Physics, University of Wisconsin - Madison, Madison, WI, United States; ²Biomedical Engineering, University of Wisconsin - Madison; ³Radiology, University of Wisconsin - Madison

Artifacts in MR often arise from data inconsistency; however, existing solutions to handle inconsistent data are often not robust, ineffective or incompatible for many applications. In this work, formulate image reconstruction to include data inconsistency from both stochastic noise and systematic bias. With k-space data inconsistencies estimated from the repetitive sampling of the center of k-space of a radial trajectory, we demonstrate improved image quality and improved noise performance in both a digital phantom and in-vivo for fast spin echo applications.