Lesion Segmentation 벤치마크
Lesion Segmentation on Anatomical Tracings of Lesions After Stroke (ATLAS)
Dice
- 2015-05-18 — U-Net: Dice 0.4606
- 2017-09-21 — 2D Dense-UNet: Dice 0.4741
- 2019-08-14 — D-UNet: Dice 0.5349
| Rank | Model | Dice | Precision | IoU | Recall | Paper | Code | Year |
|---|---|---|---|---|---|---|---|---|
| 1 | D-UNet | 0.5349 | 0.6331 | – | 0.5243 | D-UNet: a dimension-fusion U shape network for chronic stroke lesion segmentation | SZUHvern/D-UNet · SWKoreaBME/D-Unet_PyTorch · nellish123/contrib | 2019 |
| 2 | 2D Dense-UNet | 0.4741 | 0.5613 | 0.3559 | 0.4875 | H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes | xmengli999/H-DenseUNet · code-implementation1/Code9 | 2017 |
| 3 | DeepLab v3+ | 0.4609 | 0.5831 | 0.3458 | – | Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation | tensorflow/models · tensorflow/models · open-mmlab/mmsegmentation · +75 | 2018 |
| 4 | U-Net | 0.4606 | 0.5994 | 0.3447 | 0.4449 | U-Net: Convolutional Networks for Biomedical Image Segmentation | labmlai/annotated_deep_learning_paper_implementations · milesial/Pytorch-UNet · open-mmlab/mmsegmentation · +484 | 2015 |
| 5 | PSPNet | 0.3571 | 0.4769 | 0.254 | 0.3335 | Pyramid Scene Parsing Network | tensorflow/models · tensorflow/models · open-mmlab/mmsegmentation · +64 | 2016 |