3D Dynamic MRI with Homotopic l0 Minimization Reconstruction

Abstract

Recovering dynamic magnetic resonance imaging (MRI) from highly undersampled raw data is of great significance, and is an important approach for achieving both high temporal and spatial resolution in dynamic contrast enhanced imaging. In this study, a method for 3D dynamic MRI with high spatiotemporal resolution and motion insensitivity was developed, which combines golden angle variable density spiral trajectory, parallel imaging and homotopic l0 minimization-based compressed sensing. Golden angle variable density spiral trajectory, which has the advantages of high data acquisition efficiency and motion insensitivity, was used to acquire k-space data. In the reconstruction algorithm, multicoil sparsity constrain was applied in the temporal total variation domain and l0 minimization, instead of traditional l1 minimization, was adopted, which could further increase the undersampling rate. Simulations and in vivo experiments demonstrated that high spatial resolution and temporal resolution can be achieved by the proposed method while maintaining the image quality. This study also suggested the advantages of homotopic l0 norm-based 3D dynamic MRI reconstruction and its clinical potential.

Keywords

• 3d dynamic magnetic resonance imaging
• golden angle spiral
• l0 minimization
• high spatiotemporal resolution
• motion-insensitive