MOTION ARTIFACT SUPPRESSION IN MRI USING REGISTRATION OF SEGMENTED ACQUISITIONS

The elimination of motion artifacts in magnetic resonance imaging is a necessary part of its application in many areas. In this work, a general model for the origin of motion artifacts is developed and techniques based on the areas of signal processing, phase-space analysis and registration are utilized to solve this problem. Invariant tissue feature maps are generated from sub-band images that are acquired in small fractions of the imaging time. Subsequently, the relative motion across the different sub-bands is analyzed to assess the motion parameters of a rigid body motion model. Given the motion parameters, a gridding procedure is used to correct and re-sample the acquired k-space thus yielding artifact-free images. The theory and implementation algorithms of the new approach are presented and the effectiveness of the correction is demonstrated by several examples