Table 2 Influence of different data enhancement methods on model effect
From: A dynamic few-shot learning framework for medical image stream mining based on self-training
Dataset | Method | Left kidney | Right kidney | Spleen | Liver | Mean |
|---|---|---|---|---|---|---|
Abdomen-MRI | w/o augmentation | 70.82 | 76.52 | 66.02 | 71.63 | 71.25 |
Clip | 70.99 | 77.15 | 67.22 | 71.96 | 71.83 | |
Augmentation Only | 71.49 | 76.53 | 67.35 | 72.07 | 71.86 | |
ARM | 72.98 | 76.53 | 68.15 | 72.46 | 72.53 | |
Abdomen-CT | w/o augmentation | 57.74 | 54.78 | 58.36 | 69.12 | 60.01 |
Clip | 58.27 | 54.03 | 58.69 | 70.05 | 60.26 | |
Augmentation only | 61.64 | 55.81 | 58.95 | 70.66 | 61.77 | |
ARM | 61.68 | 55.49 | 60.13 | 70.94 | 62.06 |