Least-square NUFFT methods applied to 2-D and 3-D radially encoded MR image reconstruction

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)
• Image Reconstruction

Background:

• Radially encoded MRI offers motion insensitivity and reduced artifacts.
• Multidimensional MRI image reconstruction from nonuniform k-space data is challenging.

Purpose of the Study:

• Develop an efficient, high-dimensional MRI image reconstruction technique for nonuniform k-space data.
• Implement and evaluate a nonuniform fast Fourier transform (NUFFT) based reconstruction framework.

Main Methods:

• Implemented multidimensional NUFFT algorithms within a k-space image reconstruction framework.
• Compared NUFFT reconstruction against the Kaiser-Bessel (KB) gridding method.
• Applied the method to 2D and 3D radially encoded simulated and physical phantoms, and in vivo studies.

Main Results:

• NUFFT reconstruction demonstrated a superior accuracy-efficiency tradeoff compared to KB gridding, particularly with on-the-fly kernel calculation.
• Using a deapodization function, like a cosine scaling factor, improved NUFFT accuracy and speed.
• The NUFFT method proved computationally less expensive than KB gridding for 3D reconstruction.

Conclusions:

• The NUFFT-based reconstruction method provides a more accurate and efficient solution for multidimensional radially encoded MRI.
• This technique is suitable for challenging 2D and 3D in vivo applications, offering advantages over conventional gridding methods.