Update README.md

test/README.md

@@ -1,6 +1,6 @@
 ## Usage Instructions
 
-All the loss functions have been tested with [nnUNet V1](https://github.com/MIC-DKFZ/nnUNet).
+All the loss functions have been tested with the nnUNetTrainer in [nnUNet V1](https://github.com/MIC-DKFZ/nnUNet).
 
 0. Prerequisites: install [nnUNet](https://github.com/MIC-DKFZ/nnUNet).
 1. Download the loss functions: `git clone https://github.com/JunMa11/SegLoss.git`
@@ -48,5 +48,20 @@ All the loss functions have been tested with [nnUNet V1](https://github.com/MIC-
 
 ## To Do
 
-- [ ] Evaluate commonly used loss functions with [nnU-Net V2](https://github.com/MIC-DKFZ/nnUNet) because the latest version is more popular (and also better performance).
+- [ ] Evaluate commonly used loss functions with [nnU-Net V2](https://github.com/MIC-DKFZ/nnUNet) because the latest version is more popular (and has better performance).
+
+> In nnU-Net V2, deep supervision is added to the default U-Net. The optimizer is SGD with momentum rather than Adam.
+
+The associated segmentation results will be released by 4.15.
+
+| Loss | LiverTumor-DSC | LiverTumor-NSD | Pancreas-DSC | Pancreas-NSD | Multiorgan-DSC | Multiorgan-NSD |
+|------------|:--------------:|:--------------:|:------------:|:------------:|:--------------:|:--------------:|
+| CE | 0.6415 | 0.4698 | 0.8338 | 0.6566 | 0.8570 | 0.7368 |
+| Dice | 0.6200 | 0.4592 | 0.8399 | 0.6663 | 0.8577 | 0.7416 |
+| DiceCE | 0.6185 | 0.4493 | 0.8410 | 0.6691 | 0.8626 | 0.7488 |
+| DiceFocal | 0.6303 | 0.4705 | 0.8401 | 0.6691 | 0.8631 | 0.7501 |
+| DiceTopK10 | 0.6691 | 0.5095 | 0.8387 | 0.6661 | 0.8636 | 0.7483 |
+| TopK10 | 0.6512 | 0.4849 | 0.8383 | 0.6649 | 0.8560 | 0.7378 |
+
+