Table 6.
The results, both (a) average, and (b) median DICE over our clinical MRI data of low-grade glioma (WHO II) patients in the whole-tumor segmentation task, for different augmentation scenarios.
| Augmentation | Training | Validation | Test | |
|---|---|---|---|---|
| (a) | Without | 0.823 | 0.743 | 0.763 |
| Flip | 0.836 | 0.790 | 0.785 | |
| DIR | 0.858 | 0.777 | 0.773 | |
| DIR + Flip | 0.865 | 0.808 | 0.800 | |
| (b) | Without | 0.823 | 0.779 | 0.785 |
| Flip | 0.838 | 0.808 | 0.797 | |
| DIR | 0.859 | 0.802 | 0.792 | |
| DIR + Flip | 0.867 | 0.816 | 0.809 |
The results come from our paper (Nalepa et al., 2019a). The best results are boldfaced.