Model card
MedNeXt-L.
German Cancer Research Center (DKFZ)open-source62M paramsConvNeXt-based encoder-decoder2 current SOTA
ConvNeXt blocks adapted for 3D medical image segmentation. Large variant. Significant improvement over nnU-Net on multiple benchmarks.
§ 01 · Benchmarks
Every benchmark MedNeXt-L has a recorded score for.
| # | Benchmark | Area · Task | Metric | Value | Rank | Date | Source |
|---|---|---|---|---|---|---|---|
| 01 | ACDC | Medical · Medical Image Segmentation | mean-dsc | 92.7% | #1 | 2023-03-17 | source ↗ |
| 02 | BraTS 2023 | Medical · Medical Image Segmentation | mean-dice-wt-tc-et | 0.9% | #1 | 2023-03-17 | source ↗ |
| 03 | Synapse Multi-Organ CT | Medical · Medical Image Segmentation | mean-dsc | 85.5% | #2 | 2023-03-17 | source ↗ |
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Papers
1 paper with results for MedNeXt-L.
- 2023-03-17· Medical· 3 results
MedNeXt: Transformer-driven Scaling of ConvNets for Medical Image Segmentation
§ 04 · Related models
Other German Cancer Research Center (DKFZ) models scored on Codesota.
§ 05 · Sources & freshness
Where these numbers come from.
arxiv
3
results
3 of 3 rows marked verified.