MRI: reduce image noise by enforcing a data consistency constraint
MRI acquired from multiple receivers can improve the image quality by enforcing a data consistency constraint, as the image data from receivers are dependent.
Here is an example of enforcing this constraint to improve the image quality of diffusion-weighted imaging.
The zipped data can be downloaded here. Unzip the file to put all image data (*.bfloat and *.hdr) at a folder named '../meas/001' relative to the script file for implementing the data consistency constraint (see below).
Data were acquired from a 3T MRI scanner using a 32-channel head coil array. One non-diffusion-weighted image and six diffusion-weighted images (b = 1000 s/mm²) from one slice were included.
Run this script to reconstruct both non diffusion-weighted (1 image; b-value =0) and diffussion-weigted images (6 images; b-value = 1000 s/mm²; 6 different directions) with the data consistency constraint.
These are non-diffusion-weighted images without (left) and with (right) the data consistency constraint.
These are diffusion-weighted images (b-value = 1000 s/mm²) without (left) and with (right) the data consistency constraint.
