Free-breathing abdominal MR imaging for reduction of respiratory motion artifacts

Abstract

Abdominal magnetic resonance imaging (MRI) remains as a major challenge due to respiratory motions, which degrade image quality. To minimize the motion-related artifacts, free-breathing methods for retrospective reconstruction based on the respiratory self-gating have been proposed. However, in these methods, acquisitiontimings of the adjacent phase-encoding (PE) lines can be concentrated in non-quiescent respiratory phases (e.g. mid-inspiration), resulting in severe image artifacts. To overcome the problem, a robust data acquisition scheme based on a variable density sampling is presented in this thesis, where multiple signals of the same PE are equidistantly acquired over a respiratory cycle and the consecutive PE steps are randomly distributed. After gating procedure the data represent randomly undersampledpatterns along the respiratory phase direction. Thus, by using compressed sensing (CS) reconstruction technique, artifacts caused by undersampling and motion inconsistency can be effectively eliminated.To extract the respiratory motion state, the self-navigated technique is embedded in a Cartesian two dimensional fast low angle shot sequence. A variable density acquisition, whose distribution is determined by Gaussian function with an adjustable variance, is used. In the proposed data ordering, firstly, signals of the same PE step are equidistantly acquired over a predetermined respiratory period, Tresp and secondly, the ordering of the PE steps is incoherently traversed using a random generator, thereby preventing the adjacent PE lines from being densely acquired at non-quiescent respiratory phases. After retrospective gating procedure, the multiple images at different respiratory phases are reconstructed from randomly undersampled k-space data using reconstruction method with total variation as a sparsifying transform.To validate the proposed method, a numerical phantom simulation and in-vivo experiments are conducted. Simulation and experimental result...