Mass Classification on VinDr Mass
0.814AUCMIL-PF (MedSigLIP) AT: attn
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MIL-PF (MedSigLIP) AT: attn# Trainable Params (M)=0.04, Level=Breast2026.03 | 0.814 | 75.7 | 40.3 | |
| FPN-AbMIL (mean)# Trainable Params (M)=1.76, Level=Breast2026.03 | 0.808 | 75.7 | 38.7 | |
| FPN-SetTrans (mean)# Trainable Params (M)=1.76, Level=Breast2026.03 | 0.807 | 75.5 | 31.7 | |
| FPN-SetTrans (max)# Trainable Params (M)=5.38, Level=Breast2026.03 | 0.801 | 74.5 | 34.2 | |
| MIL-PF (DINOv2) AT: attn# Trainable Params (M)=0.05, Level=Breast2026.03 | 0.8 | 73.6 | 41.9 | |
| FPN-AbMIL (max)# Trainable Params (M)=5.38, Level=Breast2026.03 | 0.797 | 68.3 | 33.2 | |
| FPN-AbMIL# Trainable Params (M)=1.76, Level=Image2026.03 | 0.792 | 72 | 38.2 | |
| FPN-SetTrans# Trainable Params (M)=5.38, Level=Image2026.03 | 0.785 | 73.4 | 36.7 | |
| MIL-PF (MedSigLIP) AT: max# Trainable Params (M)=0.04, Level=Breast2026.03 | 0.778 | 72 | 40.8 | |
| GMIC-ResNet18# Trainable Params (M)=22.49, Level=Image2026.03 | 0.772 | 70.3 | 30.7 | |
| MIL-PF (DINOv2) AT: max# Trainable Params (M)=0.05, Level=Breast2026.03 | 0.77 | 73.1 | 31.1 | |
| GMIC# Trainable Params (M)=14.11, Level=Image2026.03 | 0.755 | 68.7 | 28.6 | |
| ES-Att# Trainable Params (M)=22.89, Level=Case2026.03 | 0.738 | 66.9 | 26.6 |