Table 6 Ablation Study of Feature Fusion and Knowledge Distillation (Averaged Results across Kumar, CPM-15, CPM-17, and TNBC Datasets)
From: Instance-level semantic segmentation of nuclei based on multimodal structure encoding
Method | DICE | AJI | PQ |
|---|---|---|---|
Feature fusion | |||
Single-scale features | 0.805 | 0.645 | 0.725 |
Multi-scale features | 0.825 | 0.665 | 0.745 |
Knowledge distillation | |||
Without knowledge distillation | 0.810 | 0.650 | 0.730 |
With knowledge distillation | 0.830 | 0.670 | 0.750 |