| Approach | F1-Score |
|---|---|
| Proposed Method (Self Supervision) | 92.2 |
| B6F-SVM (Classical Approach) | 89.7 |
| DenseNet (Deep Neural Network) | 89.5 |
| GCN-AF (Euclidean distance-based Deep GCN Method) | 83.8 |
| GCN-AF (Cell features-based Deep GCN Methods) | 85.8 |
| TPCD-Hist (cell-types as a feature vector) | 86.1 |