The influence of various adaptive radial undersampling schemes on compressed-sensing L1-regularized reconstruction

Introduction Adaptive imaging allows multiple image sets, each having a different spatial-temporal balance, to be retrospectively reconstructed from the same dataset. High temporal resolution image sets from radial sampling schemes are typically undersampled, and suffer from streak artifacts that degrade image quality. It has been shown that a compressed sensing (CS) L1-penalized reconstruction can be used reduce these streak artifacts [13]. However, it remains unclear which radial sampling pattern is optimal for CS reconstruction. Here, we compare the effects of 3 adaptive sampling schemes (golden angle [4-5], bit-reversed, and random radial sampling scheme) on the ability of CS reconstruction to reduce streak artifacts, at various spatiotemporal resolutions.