Faster Scanning with Parallel MRI Using a Non-Rectangular Field-Of-View

Conventional magnetic resonance imaging (MRI) uses a rectangular field-of-view (FOV) to specify the sampling pattern. The full sampling pattern is a Cartesian grid that satisfies the Nyquist-Shannon sampling theorem in each dimension of this FOV. Existing acceleration methods - such as partial Fourier sampling, parallel imaging, and compressed sensing - reduce the number of samples when compared to the fully sampled pattern. As human bodies are not rectangular, the rectangular FOV necessarily includes empty space. In this work, we show that the sampling pattern can be reduced with a non-rectangular FOV that better isolates empty space from subject anatomy. We show that this sampling pattern can be further accelerated in the parallel imaging setting where multiple coils are used for sensing. As the non-rectangular FOV could easily be provided by the technologist controlling the scanner, this method could readily be implemented on clinical scanners to yield faster scanning.