Cine CMR video generation on UKB (test)
0.27LPIPSECGFlowCMR
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ECGFlowCMRRef.=Ours, Inference time=0.45 sec / vid.2026.01 | 0.27 | 37.28 | 14.41 | |
| CardioNetsRef.=NEJM AI’26, Inference time=4.33 sec / vid.2026.01 | 0.28 | 89.83 | 21.65 | |
| EchoPulseRef.=ICLR’25, Inference time=0.84 sec / vid.2026.01 | 0.31 | 85.8 | 21.41 | |
| Cross-Modal AutoencoderRef.=Nat. Commun.’23, Inference time=0.63 sec / vid.2026.01 | 0.32 | 99.24 | 28.62 | |
| VideoGPTRef.=ArXiv’21, Inference time=2.15 sec / vid.2026.01 | 0.35 | 107.41 | 24.99 | |
| ModelScopeT2VRef.=ArXiv’23, Inference time=3.87 sec / vid.2026.01 | 0.37 | 104.19 | 31.43 |