Table 3. Discriminant function determining disease classification accuracy of multimodal MRI analyses differentiating dementia subtypes & control.
| Modalities | Model accuracy (%) | Features (#) | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) | Youden’s index (J) |
|---|---|---|---|---|---|---|---|
| All modalities | 48.8 | 49 | 73.3 | 53.9 | 47.8 | 77.8 | 27.2 |
| ASL | 45.2 | 18 | – | – | – | – | – |
| MRS | 19.0 | 10 | 33.3 | 68.8 | 28.6 | 73.3 | 2.1 |
| T1 | 42.6 | – | 53.0 | 46.7 | 36.0 | 63.6 | −0.4 |
| DTI | 40.0 | 37 | 59.0 | 47.8 | 52.0 | 55.0 | 6.9 |
Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and total model accuracy (Model accuracy) from each discriminant function analysis assessing the ability of each of the MRI modalities to differentiate patients with Alzheimer’s disease from Parkinson’s disease from the healthy control group. MRI modalities included: arterial spin labelling (ASL), magnetic resonance spectroscopy (MRS), structural T1, and diffusion tensor imaging (DTI).