Image Reconstruction of Low SNR Images from Large-N Arrays
Triantafyllou C, Wald L, Potthast A, Wiggins G
A.A. Martinos Center for Biomedical Imaging
In this work, we simulate the SNR penalty for SoS combination in low SNR images using noise maps and coil sensitivity data obtained from 8, 23, and 90channel head arrays by comparing the SoS method to combination based on coil sensitivity maps. Our findings show noise amplification of the SoS method within a low SNR object increases significantly with the number of array elements and the penalty is maximum near the array elements. For cases where coil sensitivity cases are not available, we test a modified version of the SoS which improves performance in low SNR data from large arrays.