Flux MRI: Accelerating with Aid of Physical Models

In the context of Magnetic Resonance Imaging (MRI), in vivo blood flow measurements can be obtained directly by Phase-Contrast (PC) technique. Another possibility lies in the approach by alternative flow estimation methods, such as Computational Fluid Dynamics (CFD) simulations. Despite being widely applied on exams, PC-MRI can result in excessively long scan times and may suffer from partial volume effects. Patient specific CFD simulations, on the other hand, require reduced scan times and provide high spatial and temporal resolution. At the same time, its accuracy can be hugely affected by model assumptions and simulated velocity fields can be inconsistent when compared to the subject blood flow. In order to reduce scan times, Compressive Sensing (CS) technique has been applied on flow MRI exams, such as PC-MRI, to allow blood flow to be measured from a small dataset in k-space. The main objective of this work is, under the CS framework, to propose a novel non-linear signal reconstruction method that can combine PC-MRI signal data with a priori information based on velocity images. Its feasibility is then verified, showing its capacity to integrate these distinct nature signals. Finally, it is expected, by applying the proposed approach, that patient specific blood flow measurements could be quantified by an even smaller set of measurements acquired in k-space. Then, drastically reducing examination and sampling times.