Practical Nonlinear Gradient Encoding for Enhanced Accelerated Imaging

点击翻译按钮获取中文摘要

Project Summary/Abstract This project will benefit public health by ushering in a new technology to reduce MRI scan times. FRONSAC adds nonlinear gradients of modest amplitude in dynamic waveforms to standard Cartesian MRI protocols. This small perturbation to the standard Cartesian MRI scan allows for greater undersampling of data, and faster acquisition times. This proposal is innovative because it marries the reliability of Cartesian imaging with a small amount of nonlinear gradient encoding to greatly enhance parallel imaging acceleration. The proposal is significant because it will demonstrate that a single well-characterized nonlinear gradient waveform can improve undersampled imaging for a broad range of clinical sequences and scan prescriptions, potentially doubling overall scan throughput for busy clinical sites. The aims will demonstrate that, for many scan prescriptions, and even in the presence of common experimental imperfections or other parallel imaging strategies, FRONSAC further multiplies acceleration by an additional 2-4x. 1) Sequence development and optimization. a) Develop a FRONSAC waveform, with 3 NLG channels, for a GRE sequence that maximizes clinically acceptable acceleration. b) Using this FRONSAC waveform, develop a set of widely used clinical sequences implementing FRONSAC acceleration: 3D MP-RAGE, bSSFP, TSE and T2w-FLAIR. 2) Demonstrate FRONSAC imaging in vivo. a) Acquire human brain images and compare contrast with conventional encoding. b) Compare undersampling performance between FRONSAC and Cartesian encoding. 3) Show that undersampled images acquired using a FRONSAC gradient optimized for a single geometry shows persistent improvements over Cartesian encoding for human brain imaging (for different geometries and under various common experimental imperfections), and retains compatibility with other acceleration approaches. a) Test for undersampling artifacts when changing FOV, resolution, and slice orientation. b) Test sequences with known (introduced) imperfections: gradient timing errors, imperfect shim, or off-resonance spins. c) Demonstrate compatibility with both kz undersampling and multislice CAIPIRINHA.