Chest CT Report Generation on J-MID chest CT (average across institutions)
40.08BLEU-1FedTAR
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| FedTARimage_encoder=Convolutional vision Transformer, text_decoder=DistilGPT2, adapter=LoRA (r=4)2026.02 | 40.08 | 25.94 | 17.8 | 12.4 | 29.54 | 42.8 | |
| FedAvgimage_encoder=Convolutional vision Transformer, text_decoder=DistilGPT2, adapter=LoRA (r=4)2026.02 | 38.4 | 24.68 | 16.16 | 10.98 | 28.54 | 31.7 | |
| FedProximage_encoder=Convolutional vision Transformer, text_decoder=DistilGPT2, adapter=LoRA (r=4)2026.02 | 38.32 | 24.48 | 15.95 | 11 | 28.58 | 31.62 | |
| FedAdamimage_encoder=Convolutional vision Transformer, text_decoder=DistilGPT2, adapter=LoRA (r=4)2026.02 | 38.26 | 24.54 | 16.05 | 10.93 | 28.61 | 30.62 | |
| DRFAimage_encoder=Convolutional vision Transformer, text_decoder=DistilGPT2, adapter=LoRA (r=4)2026.02 | 36.8 | 24.02 | 16.63 | 11.6 | 28.75 | 29.51 | |
| SCAFFOLDimage_encoder=Convolutional vision Transformer, text_decoder=DistilGPT2, adapter=LoRA (r=4)2026.02 | 35.4 | 22.38 | 14.8 | 10.1 | 27.12 | 24.82 | |
| FedYogiimage_encoder=Convolutional vision Transformer, text_decoder=DistilGPT2, adapter=LoRA (r=4)2026.02 | 35.39 | 22.37 | 13.72 | 10.46 | 27.91 | 24.79 |