A fast algorithm for optimal port-combination of multi-port coil in high-field MRI

A fast algorithm is proposed to choose the optimal port-combination for the next generation non-phased array coil. This kind of coil is so promising and scalable that it would mimic any known phased array coil at the cost of a combinatorial optimization problem. The raw data of signal and noise are respectively collected using a Siemens Trio 3T scanner according to the Kellman-McVeigh method. The B1 maps are acquired by the double-angle method. The object is to find the optimal port-combination which results in a high flip angle-to-SAR (specific absorption rate) ratio and SNR in the central region of coronal FOV. The number of feasible solutions to this combinatorial optimization problem raises fast with respect to the port number. Based on the positive definiteness of the noise covariance matrices, the updated Cholesky factorization is employed to accelerate the computation of SNR scaled images in the Roemer's optimal reconstruction. Numerical results show that this algorithm is much faster than the direct-inverse algorithm.