Diagnosis of schizophrenia with functional connectome data: a graph-based convolutional neural network approach

BMC Neuroscience

Table 6 Performance comparison of the BrainNet-GA CNN with competing methods

	Accuracy	Sensitivity	Specificity	AUC
SVM-PCA	74.90	77.96	71.55	78.85
SVM	72.34	76.91	67.40	76.25
FNNs	74.59	77.72	71.25	78.82
CNNs	76.79	79.65	73.68	80.69
BrainNetCNNs	77.04	78.98	75.00	81.74
SENet	81.21	83.38	79.10	86.85
BrainNet-A CNN	82.04	84.47	79.63	88.41
BrainNet-GA CNN	83.13	85.96	80.12	89.42

Data given as %, AUC, Area under the curve; BrainNet-A CNN, BrainNet-Attention CNN; BrainNet-GA CNN, BrainNet-Global Covariance Pooling-Attention CNN; CNNs, Convolutional Neural Networks; FNNs, Fully Connected Neural Networks; PCA, Principal Component Analysis; SENet, Squeeze and Excitation Network; SVM, Support Vector Machine

ISSN: 1471-2202