Predict Alzheimer’s disease using hippocampus MRI data: a lightweight 3D deep convolutional network model with visual and global shape representations

Table 3 Performance comparison of DenseCNN2 with other methods

Method	Accuracy	Sensitivity	Specificity	AUC
Hippo volumes	50.54	42.12	58.40	53.54
Hippo CHF features	85.12	76.31	81.40	–
3D CNN	86.94	79.36	93.21	86.40
ResNet	90.00	–	–	95.60
Hybrid CNN-RNN	89.17	84.64	93.16	91.00
Multi-model CNN	88.90	86.62	90.81	92.50
3D DenseNet	92.29	90.63	93.72	96.95
DenseCNN2	92.52	88.20	94.95	97.89

ISSN: 1758-9193