Classification of 12-lead ECGs: the PhysioNet/Computing in Cardiology Challenge 2020
Digest of Top Models (CinC2020 papers searching Challenge
sessions on this page)
- prna: Transformer (+ CNN + Manual (statitical) features)
- Between a ROC and a heart place: Adaptive lead weighted ResNet
- HeartBeats: SE-ResNet
- Triage: ResNet and Scatter Transform
- Sharif AI Team: Branched CNN (each branch being VGG-like) and RNN
- DSAIL_SNU: Bag of Tricks (data augmentation?) for Electrocardiogram Classification with DNN
- UMCUVA: Exponentially Dilated Causal CNN
- CQUPT_ECG: Multi-scale SE-Net
- ECU: CRNN + NAS
- PALab: SE-ResNet
- BUTTeam: CNN (modified resnet) With Global Skip Connections and Custom Loss Function
- SpaceOn Flattop: Modified (how modified?) ResNet
- ISIBrno: Residual CNN-GRU with Attention Mechanism
- other teams: Modified EfficientNet; Class Activation Map; Graph Convolutional Network (GCN); etc.
Final rankings can be found here