Significance of Normalization on Anatomical MRI Measures in Predicting Alzheimer’s Disease
Table 3
Classification performances on raw data.
Model
Accuracy
Sensitivity
Specificity
Precision
MMSE
88.3
(87.3–89.4)
81.0
(76.7–81.7)
91.6
(91.5–94.6)
82.6
(81.3–87.6)
Subcortical volume (SV)
83.1
(81.5–85.2)
77.9
(75.0–80.0)
85.6
(83.0–88.4)
72.6
(69.2–77.6)
Cortical thickness (CT)
77.7
(76.2–78.9)
74.8
(73.3–76.7)
79.0
(77.4–80.7)
63.0
(59.9–68.0)
Surface area (SA)
71.4
(68.3–73.6)
58.7
(53.3–65.0)
77.2
(73.6–79.8)
55.0
(51.9–58.9)
Average
80.1
73.1
83.4
68.3
Hierarchical model
MMSE + SV
92.3
(90.5–93.1)
88.2
(85.0–90.7)
94.2
(92.3–95.3)
88.3
(85.1–90.5)
MMSE + CT
91.4
(90.4–92.6)
85.3
(83.3–88.3)
94.2
(93.0–95.4)
87.8
(85.1–90.3)
MMSE + SA
88.6
(86.3–89.5)
76.3
(71.7–78.3)
94.3
(91.5–95.4)
87.1
(81.4–89.9)
CT + SV*
83.1
(81.5–85.2)
77.9
(75.0–80.0)
85.6
(83.0–88.4)
72.6
(69.2–77.6)
SA + CT + SV*
83.1
(81.5–85.2)
77.9
(75.0–80.0)
85.6
(83.0–88.4)
72.6
(69.2–77.6)
MMSE + SA + CT + SV**
92.3
(90.5–93.1)
88.2
(85.0–90.7)
94.2
(92.3–95.3)
88.3
(85.1–90.5)
Average
88.5
82.3
91.4
82.8
The results of these models are the same as those of model of “SV” since the variables extracted for the decisional space are the same as those for “SV” model.
**This model gives identical results as those of the model of “MMSE + SV” since variables extracted for the decisional space are the same as those for “MMSE + SV.”
