Christopher Thomas Rodgers1, Stefan Neubauer1, Matthew D. Robson1
1Oxford Centre for Clinical Magnetic Resonance Research, University of Oxford, Oxford, UK
Array receive coils are ubiquitous for magnetic resonance (MR) imaging, improving the signal-to-noise ratio (SNR) and field of view. For MR spectroscopy, array coils are less well established and it is less obvious how signals from each element should be recombined. We present a Bayesian model of array spectroscopy where the maximum likelihood spectrum can be recovered using the well-known singular value decomposition. This simple and efficient algorithm is compatible with 1H and heteronuclear spectroscopy and does not necessitate tedious curve fitting procedures. We use 31P cardiac spectra from an eight-element array and numerical simulations to demonstrate its effectiveness.