In NMR abdominal imaging, motion artifacts such as ghosts and blurs degrade the image. As a motion-artifact reduction method, ROPE (Respiratory Ordered Phase-Encoding) technique has been previously developed. ROPE is known to be efficient in eliminating ghosts, however, it can not completely remove blurs. Recently, inversely weighted reordering scheme for reduction of both ghosts and blurs has been reported. These techniques using the concept of reordering to reduce motion-artifacts require too many data sets to correct motion-artifacts, i.e., long data acquisition times. In this thesis, sinusoidal scan in k-space that is a kind of modulated phase-encoding gradient aiming at a reduction of data acquisition time is suggested as an alternative to normal KWE(Kumar-Welti-Earnst) method. And the proposed method is applied to reduction of motion-artifacts with both ROPE and the inversely weighted reordering scheme. Through the theoretical analysis, computer simulations, and experiments it will be shown that the proposed method provides faster imaging and more motion-insensitive reordering effect than normal KWE method. Reduction ratio of data acquisition time by a factor of four is attainable in practice. Experiments are performed with KAIST 2.0T superconducting magnet NMR imaging system using water-contained phantom.