Medical Image Segmentation on AMOS CT
85.21Dice CoefficientK-Prism
Evaluation Results
| Method | Links | |
|---|---|---|
| K-PrismModel Category=Universal Model, Evaluation Protocol=Semantic2025.09 | 85.21 | |
| SegMoTEinteraction_type=bounding-box, trainable_params=17M, decoder_status=unfrozen2026.02 | 85.16 | |
| HermesModel Category=Universal Model, Evaluation Protocol=Semantic2025.09 | 85.06 | |
| UniSegModel Category=Universal Model, Evaluation Protocol=Semantic2025.09 | 84.91 | |
| Clip-drivenModel Category=Universal Model, Evaluation Protocol=Semantic2025.09 | 84.37 | |
| IMISinteraction_type=bounding-box2026.02 | 83.56 | |
| Transfer-AwarePretraining Strategy=Data mixture (T = 100K)2026.05 | 83.1 | |
| Data-ProportionalPretraining Strategy=Data mixture (T = 100K)2026.05 | 81.6 | |
| SAM-Med2Dinteraction_type=bounding-box2026.02 | 80.71 | |
| K-PrismEvaluation Protocol=In-context, Inference Protocol=one-shot2025.09 | 79.72 | |
| Single-Modality onlyPretraining Strategy=No mixture2026.05 | 79.3 | |
| SAM2interaction_type=bounding-box2026.02 | 78.84 | |
| nnU-NetModel Category=Task-specific Model, Evaluation Protocol=Semantic2025.09 | 76.26 | |
| SAMinteraction_type=bounding-box2026.02 | 76.05 | |
| From scratchPretraining Strategy=No mixture2026.05 | 76 | |
| IrisEvaluation Protocol=In-context, Inference Protocol=one-shot2025.09 | 74.1 | |
| TycheEvaluation Protocol=In-context, Inference Protocol=one-shot2025.09 | 61.62 | |
| UniverSegEvaluation Protocol=In-context, Inference Protocol=one-shot2025.09 | 60.51 | |
| MultiverSegEvaluation Protocol=In-context, Inference Protocol=one-shot2025.09 | 58.48 |