Table 4.
Classification performance on the testing groups for considered algorithms under evaluation criteria. Top: baseline approaches. Bottom: ANN pretrainings.
| Algorithm | α | τ NC | τ MCI | τ AD | β | β NC | β MCI | β AD |
|---|---|---|---|---|---|---|---|---|
| 2014 CADDementia | ||||||||
| Sørensen | 63.0 | 96.9 | 28.7 | 61.2 | 78.8 | 86.3 | 63.1 | 87.5 |
| Wachinger | 59.0 | 72.1 | 51.6 | 51.5 | 77.0 | 83.3 | 59.4 | 88.2 |
| Ledig | 57.9 | 89.1 | 41.0 | 38.8 | 76.7 | 86.6 | 59.7 | 84.9 |
| Abdulkadir | 53.7 | 45.7 | 65.6 | 49.5 | 77.7 | 85.6 | 59.9 | 86.7 |
|
| ||||||||
| “best” quality testing | ||||||||
| NN-AEN | 47.6 | 73.4 | 33.1 | 38.1 | 64.9 | 71.4 | 53.4 | 75.1 |
| NN-PCA | 63.8 | 70.4 | 56.7 | 66.9 | 80.0 | 87.2 | 70.0 | 87.0 |
| NN-CKA | 70.9 | 78.4 | 66.6 | 68.3 | 85.3 | 91.7 | 78.4 | 88.3 |
|
| ||||||||
| “partial” quality | ||||||||
| NN-AEN | 62.9 | 64.6 | 46.4 | 32.0 | 77.0 | 82.5 | 65.6 | 72.5 |
| NN-PCA | 64.4 | 67.6 | 49.3 | 26.0 | 78.4 | 82.3 | 67.5 | 79.2 |
| NN-CKA | 65.2 | 68.6 | 38.6 | 42.0 | 81.6 | 85.7 | 70.1 | 82.4 |