Figure 2
From: Deep learning phase error correction for cerebrovascular 4D flow MRI
Diagram shows strategy for inferring background phase correction from input uncorrected 4D Flow MRI data. Uncorrected 4D Flow MRI velocity and magnitude data were used as standardized volumes to train a modified 3D UNet to infer pixelwise background phase correction. Post-processing entailed 3rd order polynomial regression of 3D UNet inferences.
