Abstract
Magnetic resonance imaging (MRI) applications to the study of gastric function in humans have started to incorporate dynamic volumetric imaging, thus calling for specialized approaches for motion correction. A method for retrospective respiratory motion correction in free-breathing, four-dimensional (4D) abdominal MRI is presented. Our gastric low-rank tensor-based (GLOW) algorithm uses a low-rank tensor (LRT) model to separate the temporal components that correspond to breathing motion from those related to gut motion, which are preserved due to being uncorrelated and spatially localized. As a proof-of-concept, the GLOW algorithm is applied to a human 4D gastric MRI dataset that includes data collected during both a fasted and fed state using a food-based contrast meal. This approach allows for a more robust and accurate assessment of gastric peristalsis. The GLOW algorithm represents an important step toward the effective application of noninvasive, naturalistic approaches to robustly and accurately evaluate gastric function via MRI.