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

Table 9

Ablations studies of the proposed 3DCNN + SVM model applied to binary classification of AD vs. MCI (%).

Method	Training set				Testing set
	ACC	SEN	SPE	AUC	ACC	SEN	SPE	AUC
3DCNN	98.37	99.73	97.72	99.81	73.56	73.84	73.51	77.82
3DCNN + SVM	97.72	99.43	96.80	99.84	73.95	71.88	74.89	78.75
3DCNN + SVM + E2E	98.45	99.24	97.31	99.91	74.29	70.78	75.48	80.11