Skip to content

Improving high frequency image features of Deep Learning reconstructions via k-space refinement with null-space kernel

Notifications You must be signed in to change notification settings

KHRyu8985/RefineDLMRI

Repository files navigation

RefineDLMRI

Improving high frequency image features of Deep Learning reconstructions via k-space refinement with null-space kernel

Accepted in Magnetic Resonance in Medicine (2022)

March 19. 2022 Kanghyun Ryu (kanghyun@stanford.edu)

This repo contains code for Refining DL reconstruction via null-space kernel.

Refinement

As can be seen in the Figure, (a) is undersampled MRI reconstruction from UNN (Unrolled Neural Network) and (b) is the proposed refinement process.

By utilizing a null-space kernel, we can correct for errors in Deep Learning's kspace estimates and refine reconstruction.

Please refer to demo.ipynb for demonstration of running the code and examples.

About

Improving high frequency image features of Deep Learning reconstructions via k-space refinement with null-space kernel

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published