Table 2.
Correlation analysis between the T1-weighted MR image-based features, T2-weighted MR image-based features and multi-level fusion-based features
| Class | T1 image | T2 image |
|---|---|---|
| PET image | 6 | 8 |
| Matrix-level fusion | 81 | 72 |
| Feature-level fusion | 68 | 71 |
| Image-level fusion (0.1) | 10 | 8 |
| Image-level fusion (0.2) | 11 | 8 |
| Image-level fusion (0.3) | 11 | 11 |
| Image-level fusion (0.4) | 11 | 14 |
| Image-level fusion (0.5) | 13 | 21 |
| Image-level fusion (0.6) | 14 | 56 |
| Image-level fusion (0.7) | 15 | 70 |
| Image-level fusion (0.8) | 48 | 76 |
| Image-level fusion (0.9) | 64 | 87 |
For the class column, the number in the parentheses indicated the fusion weight of the MR images. The “matrix-level fusion” indicated the T1/PET matrix fusion-based features for the correlation analysis with T1-weighted MR-based features, and the T2/PET matrix fusion-based features for the correlation analysis with T2-weighted MR-based features. The “feature-level fusion” indicated the feature average method based fusion