Motion Detection and Correction in Magnetic Resonance Imaging

Magnetic resonance imaging (MRI) is a non-invasive technique used to produce high-quality images of the interior of the human body. Compared to other imaging modalities, however, MRI requires a relatively long data acquisition time to form an image. Patients often have difficulty staying still during this period. This is problematic as motion produces artifacts in the image. This thesis explores the methods of imaging a moving object using MRI. Testing is performed using simulations, a moving phantom, and human subjects. Several strategies developed to avoid motion artifact problems are presented. Emphasis is placed on techniques that provide motion correction without penalty in terms of acquisition time. The most significant contribution presented is the development and assessment of the 'TRELLIS' pulse sequence and reconstruction algorithm. TRELLIS is a unique approach to motion correction in MRI. Orthogonal overlapping strips fill k-space and phase-encode and frequency-encode directions are alternated such that the frequency-encode direction always runs lengthwise along each strip. The overlap between pairs of orthogonal strips is used for signal averaging and to produce a system of equations that, when solved, quantifies the rotational and translational motion of the object. Acquired data is then corrected using this motion estimation. The advantage of TRELLIS over existing techniques is that k-space is sampled uniformly and all collected data is used for both motion detection and image reconstruction. This thesis presents a number of other contributions: a proposed means of motion correction using parallel imaging; an extension to the phase-correlation method for determining displacement between two objects; a metric to quantify the level of motion artifacts; a moving phantom; a physical version of the ubiquitous Shepp-Logan head phantom; a motion resistant data acquisition technique; and a means of correcting for T2 blurring artifacts