Research Article
Significance of Normalization on Anatomical MRI Measures in Predicting Alzheimer’s Disease
Table 4
Classification performances on normalized data.
| Model | Accuracy | Sensitivity | Specificity | Precision |
| Subcortical volume (SV) | 83.5 | (82.0–84.7) | 74.4 | (71.7–76.7) | 87.7 | (95.3–90.0) | 75.2 | (72.0–79.0) | Cortical thickness (CT) | 79.0 | (77.8–80.4) | 78.8 | (75.0–81.7) | 79.2 | (77.5–80.6) | 64.5 | (61.8–87.4) | CT (Mean)* | 78.9 | (77.2–80.5) | 78.4 | (75.0–81.7) | 79.2 | (75.7–80.7) | 64.6 | (60.9–68.4) | Surface area (SA) | 72.3 | (68.8–75.2) | 42.6 | (35.0–48.3) | 86.1 | (82.1–89.2) | 60.4 | (50.8–65.3) | SA (Area)** | 72.6 | (70.3–75.1) | 61.2 | (58.3–63.3) | 77.9 | (75.1–81.4) | 57.4 | (52.9–61.8) | Average | 77.3 | 67.1 | 82.02 | 64.4 | |
| Hierarchical model |
| MMSE + SV | 91.7 | (90.0–93.1) | 85.8 | (81.7–88.3) | 94.5 | (93.1–95.4) | 88.2 | (85.0–90.7) | MMSE + CT | 91.5 | (89.4–93.2) | 86.9 | (81.7–90.0) | 93.6 | (90.8–96.1) | 87.2 | (82.7–90.9) | MMSE + CT (Mean)* | 90.3 | (89.2–91.1) | 90.8 | (89.6–91.7) | 90.1 | (88.2–91.7) | 81.4 | (78.5–83.9) | MMSE + SA | 88.3 | (87.3–88.9) | 80.9 | (76.7–81.7) | 91.7 | (91.4–93.8) | 82.7 | (81.1–86.1) | MMSE + SA (Area)** | 88.6 | (86.8–89.9) | 80.9 | (75.0–78.3) | 94.2 | (92.2–95.4) | 86.9 | (84.6–89.5) | CT + SV | 83.1 | (80.9–84.2) | 75.8 | (73.3–76.7) | 86.5 | (84.4–88.4) | 73.3 | (70.2–76.8) | CT + SA + SV | 83.4 | (81.0–85.7) | 78.0 | (75.0–80.0) | 85.9 | (83.0–89.1) | 73.2 | (68.3–69.4) | MMSE CT + SA + SV | 91.7 | (90.4–92.6) | 86.0 | (83.3–90.0) | 94.4 | (93.7–95.4) | 88.4 | (86.5–90.2) | Average | 88.6 | | 83.1 | | 91.3 | | 82.7 | |
|
|
Scaled by the mean thickness of the all the thickness measures.
**Scaled by the total area of the all the measures.
|