GLOW: Gastric LOW-Rank Tensor-Based Motion Correction for Abdominal 4D MRI

Area of Science:

• Medical Imaging
• Gastroenterology
• Biomedical Engineering

Background:

• Dynamic volumetric imaging in Magnetic Resonance Imaging (MRI) is increasingly used for studying human gastric function.
• Existing methods require specialized approaches for motion correction, particularly for respiratory motion in free-breathing, four-dimensional (4D) abdominal MRI.
• Accurate assessment of gastric motility and function is crucial for diagnosing and managing gastrointestinal disorders.

Purpose of the Study:

• To present a novel method for retrospective respiratory motion correction in 4D abdominal MRI.
• To introduce the gastric low-rank tensor-based (GLOW) algorithm for separating respiratory and gastric motion.
• To demonstrate the feasibility and effectiveness of the GLOW algorithm in human gastric MRI datasets.

Main Methods:

• Development of the gastric low-rank tensor-based (GLOW) algorithm utilizing a low-rank tensor (LRT) model.
• Application of the GLOW algorithm to human 4D gastric MRI data acquired during fasted and fed states.
• Separation of temporal components corresponding to breathing motion from those related to gut motion, preserving uncorrelated and spatially localized gut motion signals.

Main Results:

• The GLOW algorithm successfully separated respiratory motion from gastric and gut motion in 4D abdominal MRI.
• The algorithm demonstrated proof-of-concept on human gastric MRI data, including scans with a food-based contrast meal.
• Preservation of uncorrelated and spatially localized gut motion components allowed for improved analysis.

Conclusions:

• The GLOW algorithm provides a robust and accurate method for respiratory motion correction in 4D gastric MRI.
• This technique enables a more precise assessment of gastric peristalsis and function.
• The GLOW algorithm advances the application of noninvasive, naturalistic MRI techniques for evaluating gastric physiology.