Parallel imaging for hyperpolarized metabolic magnetic resonance imaging

Abstract

Hyperpolarized 13C magnetic resonance imaging (MRI) is a novel and safe technique to image in vivo metabolism. The technique relies on intravenous injection of hyperpolarized biological substrates provided through dissolution dynamic nuclear polarization. The polarization process increases the magnetic resonance (MR) signal of the substrate and metabolic products considerably and hereby enables real-time assessment of metabolism that is otherwise undetectable. Changes in normal metabolism are connected to several diseases, and hyperpolarized 13C MRI hereby represents an opportunity to better understand  these changes, for early diagnosis, and for faster treatment assessment. The technique is not limited to a certain anatomy or pathology, but has a broad clinical potential. Uptake and metabolic conversion of the injected substrate is tracked by MRI, utilizing its ability to differentiate between molecules with di#erent magnetic properties. However, there are limits to how fast data can be collected using traditional acquisition methods. This together with the fact that the high magnetization of the hyperpolarized substrate disappears on a time-scale of a few minutes represent one of the major challenges in the clinical translation of hyperpolarized 13C MRI. One method to increase MR scan e$ciency is parallel imaging, which uses sensitivity information characteristic of multi-channel receive coils to accelerate acquisition. Parallel imaging is standard in conventional MRI and has a huge potential for use with hyperpolarized 13C MRI. To investigate this potential and to suggest solutions to specific implementation challenges, multiple studies were carried out. All experiments were performed at 3 T magnetic field strength using a human clinical MR scanner to facilitate clinical translation. Three main objectives were pursued: characterizing multi-channel 13C receive coils, investigating the optimal approach to coil sensitivity calibration, and developing and testing 3D accelerated methods to parallel imaging acquisition in vivo. Coil characterization was performed through simulations and phantom experiments, while the two other objectives also involved acquisition of in vivo data that were predominantly acquired for healthy pigs with imaging of kidneys and heart. An abdominal imaging dataset from a healthy human volunteer was also collected. Methods developed and results obtained from the coil characterization study provided directions for future 13C coil design. The study investigating calibration of coil sensitivities found that pre-calibration of the sensitivities was both feasible and advantageous for parallel imaging acquisition using a multi-channel coil with fixed geometry. The final tests of 3D accelerated acquisition were first performed for healthy pigs using a pre-calibrated parallel imaging scheme, which demonstrated increased information output through higher spatial and temporal resolution of metabolite images compared to non-accelerated acquisition. Next, using a calibrationless parallel imaging scheme, the first full volumetric coverage of human abdominal metabolism was demonstrated in combination with a multi-channel coil with adjustable geometry. All in all, the research presented in this thesis clarifies underlying prerequisites and demonstrates successful implementation of parallel imaging for hyperpolarized 13C MRI.