Automatic Diagnosis of Alzheimer’s Disease and Mild Cognitive Impairment Based on CNN + SVM Networks with End-to-End Training

Table 4

Evaluation of the proposed 3DCNN + SVM with E2E applied to binary classification of MCI vs. NC samples (%).

Method	Training set				Testing set
	ACC	SEN	SPE	AUC	ACC	SEN	SPE	AUC
Gray [6]	—	—	—	—	70.20	73.80	62.30	73.0
Lu [10]	—	—	—	—	79.63	—	—	—
Silveira [8]	—	—	—	—	70.00	46.96	80.44	—
Ding et al. [12]	98.70	98.05	99.55	99.43	72.37	74.70	69.31	79.19
Liu et al. [14]	99.04	98.52	99.74	99.73	73.80	73.16	74.69	80.45
Huang et al. [15]	98.30	97.72	99.09	99.97	73.52	75.50	70.90	79.65
Proposed	99.54	99.26	99.90	99.88	76.68	77.80	75.57	82.39