- User Since
- Aug 1 2016, 12:10 PM (128 w, 6 d)
Nov 28 2018
As far as I can tell files were split along t with fslsplit.
Nov 27 2018
Nov 26 2018
I accidentally uploaded the segmentation. This is the raw data. The problem occurs with both.
Oct 26 2018
I think this is either a duplicate or this is is not correct. Please ignore
Jul 2 2018
Oversampling foreground makes the dice believe the foreground is more frequent than it actually is. Therefore, it does not punish false positives hard enough. I therefore tried to balance the dice by doing the following:
Similar to Jakobs segmentation method. 3 UNets, each trained with a different perspective. Little improvement on Hippocampus, no difference on BraTS, but worse performance on other datasets -> Max be due to anisotropy