A question about undersampling mask in singlecoil-knee dataset #250
-
The original U-net network you posted was applied 4x Random Mask to train and evaluate.But I change the mask size to only random sample a certain location but still in 4x Random Mask. Then I made some adjustment to the undersampled data in file( UnetDataTransform), and ultimately improve the metrics. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
Hello @1025754824, there have been many improvements over the original method, as you can see on the fastMRI leaderboard https://fastmri.org/leaderboards/. In terms of convincible - I'm a little less able to help there. You might be able to open some general discussion here or email some researchers in the field. |
Beta Was this translation helpful? Give feedback.
Hello @1025754824, there have been many improvements over the original method, as you can see on the fastMRI leaderboard https://fastmri.org/leaderboards/.
In terms of convincible - I'm a little less able to help there. You might be able to open some general discussion here or email some researchers in the field.