Table 5.
Performance benchmarking of our approach against leading techniques on different components across ADNI and BraTS datasets.
| Model | ADNI dataset | BraTS dataset | ||||||
|---|---|---|---|---|---|---|---|---|
| Mean average precision | Precision | Recall | F1 score | Mean average precision | Precision | Recall | F1 score | |
| w/o Unified multi-modal encoding | 85.14 0.02 | 82.05 0.03 | 83.89 0.02 | 83.32 0.02 | 84.42 0.02 | 81.78 0.02 | 83.50 0.03 | 82.67 0.02 |
| w/o Entropy-guided uncertainty modeling | 86.65 0.03 | 84.02 0.02 | 85.54 0.02 | 84.87 0.03 | 85.28 0.02 | 83.21 0.03 | 84.89 0.02 | 84.03 0.02 |
| w/o Contextual prediction adjustment | 87.39 0.02 | 85.76 0.03 | 86.92 0.02 | 86.15 0.02 | 86.09 0.02 | 84.88 0.02 | 85.99 0.03 | 85.24 0.02 |
| Ours | 90.23 0.02 | 87.89 0.02 | 89.45 0.03 | 88.76 0.02 | 88.94 0.02 | 86.45 0.02 | 87.92 0.03 | 87.21 0.02 |
The values in bold are the best values.