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.
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.