Swin UNETR is a 3D segmentation model that uses hierarchical Swin Transformer blocks as the encod...
64
Swin UNETR is a 3D segmentation model that uses hierarchical Swin Transformer blocks as the encoder of a UNETR-style network. It processes 96x96x96 patches of abdominal CT scans to segment 13 organs (spleen, kidneys, gallbladder, esophagus, liver, stomach, aorta, IVC, portal/splenic veins, pancreas, adrenal glands) plus background. Trained on the BTCV multi-atlas challenge dataset.
| Vendor | AMD |
| License | Apache-2.0 |
| Source | https://github.com/AMD-SiloAI/aim-build |
| Revision | 5f3ecefe7506f893232c83950f2afb4147804869 |
| Built | 2026-10-30T09:19:37Z |
Content type
Image
Digest
sha256:f5361450b…
Size
10.5 GB
Last updated
9 days ago
docker pull amdenterpriseai/aim-instinct-model-monai-swinunetr:2026.14.0