Hyperpolarized Xenon Imaging with the SWIFT Approach in Ultra-low Field MRI: A Simulation Study

Yuki Kaga, Takenori Oida, Tetsuo Kobayashi
Vol. 4 (2015) p. 42-47

Recently, ultra-low field MRI (ULF-MRI) has attracted attention as a medical imaging technique. In ULF-MRI, because nuclear magnetization is very weak, hyperpolarized nuclear magnetization is expected to increase the intensity of MR signals. The use of hyperpolarized xenon gas, which has substantially larger polarization ratio than protons, has been proposed. However, MR signals decay rapidly when a large gradient field is used in MR signal detection. Meanwhile, the sweep imaging with Fourier transformation (SWIFT) approach, which utilizes an adiabatic radio frequency pulse with amplitude and frequency modulations and small gradient field, has been proposed. In this study, we evaluated the influence of the parameters used in the SWIFT approach on hyperpolarized xenon imaging in ULF-MRI. To simulate signal reconstruction of hyperpolarized xenon with the SWIFT approach, we calculated motions of magnetization according to the Bloch equation using a fourth-order Runge–Kutta method. Two pyramidal profiles with widths of 25 and 75 mm were simulated as nuclear magnetization density profiles. The motions of magnetization and the acquired MR signals caused by magnetization were subsequently computed using various parameters comprising sampling points, bandwidth, and duration of excitation. The signals were finally reconstructed using a cross-correlation method. The results indicated that reconstructed signals could be calculated from the MR signals acquired by the SWIFT approach at a bandwidth of 1, 10 or 100 kHz and a matrix size of 100, 200 or 400. However, signals reconstructed using a bandwidth of 1 kHz or a matrix size of 100 were distorted near the position where the signal changed. These results suggest that a wide bandwidth (≥10 kHz) and a large matrix size (≥200) should be used for better signal reconstruction from the MR signals acquired by the SWIFT approach, and that the Larmor frequency of ULF-MRI should be greater than 10 kHz to achieve wide bandwidth. In addition, DC component distortions in the reconstructed signals increased when distribution of nuclear magnetization density in the field-of-view (FOV) was large. Therefore, a wide FOV should be selected to reduce DC component distortions.