Abstract #3655

Huang F, Yan R, Akao J, Duensing R, Saylor C, Cheng H

Variable density (vd) parallel imaging provides freedom of optimizing k-space trajectory either to increase reconstruction SNR or reduce low frequency aliasing. The reconstructions using variable density such as SPACE RIP, Generalized SMASH, and Generalized SENSE consider this problem globally by solving huge matrix equations, which is time consuming. Parallel processing to different acceleration factor regions, such as Madores method and later Kings method, reduces the reconstruction time but can have either a ringing artifact at edges or bias. This paper discusses a filter design strategy in vdSENSE to reduce bias while still filtering out the ring effect, named as homoSENSE.