Resolution evaluation of MR images reconstructed by iterative thresholding algorithms for compressed sensing

Area of Science:

• Medical Imaging
• Image Reconstruction
• Signal Processing

Background:

• Traditional magnetic resonance imaging (MRI) systems rely on linear, stationary transforms, allowing quality assessment via point spread function (PSF).
• Compressed sensing (CS) in MRI utilizes nonlinear, nonstationary reconstruction, rendering classical PSF evaluation inadequate.

Purpose of the Study:

• To adapt and apply the local point spread function (PSF) concept for assessing spatial and temporal resolution in MRI.
• To evaluate images reconstructed using iterative soft thresholding (IST) for CS MRI.

Main Methods:

• Applied a local PSF concept to assess quality of CS MRI images reconstructed with IST.
• Investigated the impact of k-space sampling patterns and sparsifying transforms on local spatial resolution.
• Determined local temporal resolution in image series reconstructed using spatiotemporal sparsity.

Main Results:

• K-space sampling and sparsifying transforms significantly impacted local image resolution.
• Reconstructions using x-f-space showed slight temporal blurring in dynamic regions.
• Local resolution demonstrated a dependence on the thresholding parameter, enabling optimization.

Conclusions:

• Local PSFs effectively evaluate local spatial and temporal resolution in CS MRI reconstructed with nonlinear algorithms.
• This method facilitates the selection of optimal acquisition parameters by analyzing the influence of thresholding and sampling.
• The approach improves MRI results through informed parameter selection.