Meeting Banner
Abstract #3352

A Quantitative Study of Sodicksons Paradox

Yu Ding1, Hui Xue2, Ti-chiun Chang3, Christoph Guetter3, Orlando Simonetti1

1Dorothy M. Davis Heart and Lung Research Institute, The Ohio State University, Columbus, OH, United States; 2siemens corporate research; 3Siemens Corporate Research

GRAPPA reconstructs the missing k-space by applying a convolution kernel which is estimated from ACS lines using linear regression. Intuitively, ACS lines with higher SNR should boost the accuracy of the kernel estimation, and increase the SNR of GRAPPA reconstruction. Paradoxically, Sodickson and his colleagues pointed out that the higher SNR in ACS lines may lead to lower SNR in GRAPPA reconstruction. We study quantitative about how the noise in the ACS lines affects the SNR of the GRAPPA reconstruction, and proposes a simple solution to improve the SNR of TGRAPPA.

Abstract PDF E-Poster Video

Keywords

abstract acceleration accuracy acquisition affecting affects amplifies amplify applying array auto averaging avoid axis bandwidth better boost calibration cardiac changing channel cine clinically coil colleagues condition convolution corporate corresponds counter degradation ding dynamic encoding equation error estimation experiment frame frames frequency full fully healthy heart horizontal hypothesis hypothesize improve improved improves in vivo indicate instead institute intuitively kernel kernels last lead linear linearly long lung magnitude matrix measured missing moderate noise optimal paradox paradoxically parallel phantom pixel please prepared preserve proportional propose proposes protocol quantitative random reaches real reconstructed reconstruction reconstructions reconstructs refer regression regularization relation relevant respectively reveals robustness root sampled scanner series sets sharpness short simple slice solution space spans spot square statistical strategy sufficient sweet thick tile tiling together trio variant varied varying versa vertical vice views volunteer widely