Fast acquisition and metabolic modelling for deuterium MRI

The human brain is the organ with the highest energy consumption relative to its weight. This energy is mostly consumed in the form of glucose, which can be metabolised through oxidative, or non- oxidative pathways. In the oxidative pathway glutamate plays a crucial role, while in the latter lactate is of importance. In several diseases such as in brain tumours or epilepsy this energy metabolism is altered, for example in the form of an altered balance between the oxidative and non-oxidative pathway. Therefore, studying the brain energy metabolism can provide valuable insight to understand and detect diseases. Magnetic Resonance Spectroscopic Imaging (MRSI) is a method to investigate the concentrations of specific metabolites using an MR scanner. A special form of MRSI is deuterium-MRSI, where only signal from the hydrogen isotope deuterium can be detected. Before consuming deuterated glucose which has some of its hydrogen atoms replaced by deuterium, no glutamate and lactate signals are detected with deuterium-MRSI, since natural glutamate and lactate have almost no deuterium. After consuming deuterated glucose, it is metabolised into glutamate and lactate, and thus signals show up during the experiment. This allows to quantify the metabolic rates of the oxidative and non-oxidative pathways. Unfortunately, the deuterium-MRSI methodology is not yet fully established, and lacks many developments that have been achieved in MRI or conventional MRSI. Therefore, the goal of this project is to improve the methods of deuterium-MRSI at a human as well as an animal MRI scanner. This includes developing a model for quantifying the oxidative and non-oxidative metabolic rates of glucose, as well as improvements in the acquisition and reconstruction. The goal is to make the acquisition faster and to reduce noise from the data. This can be achieved by measuring the data in concentric rings instead of individual points in the measurement space, and by applying a certain model assuming low-rankedness of the data to distinguish between signal and noise. Further, these developed methods should then be applied as a proof of principle to epilepsy patients and rat models of epilepsy. This will allow us to investigate the benefits of the developed methodology in a real application, and may help to better understand epilepsy. The results will be compared to the standard method for assessing the energy metabolism in epilepsy, FDG-PET.