Chapter 45. ℓ1 Minimization in ℓ1-SPIRiT Compressed Sensing MRI Reconstruction
Mark Murphy and Miki Lustig
Magnetic resonance imaging (MRI) is a rich, multidisciplinary topic whose modern success is due to a synthesis of knowledge from many fields: medicine, physics, chemistry, engineering, and computer science. This chapter discusses the successful Cuda parallelization of an algorithm (ℓ 1-SPIRiT
[5]
: ℓ 1-regularized
iT
erative
S
elf-consistent
P
arallel
I
maging
R
econstruction) for reconstructing MR images acquired using two advanced methods: compressed sensing and parallel imaging (CS+PI).
The issues discussed in this chapter are simlilar to those faced by many MRI and statistical signal processing applications. Computationally, ℓ 1-SPIRiT ...