TABLE 2.

Classification performance of different methods on multi-centers dataset.

Method	Accuracy (%)	Sensitivity (%)	Specificity (%)	Precision (%)	Running time(s)
SVM	60.3 ± 3.6	35.3 ± 8.2	84.4 ± 6.6	64.3 ± 5.9	186
Random forest	63.7 ± 7.2	54.9 ± 3.9	71.3 ± 9.2	68.9 ± 3.8	67
DNN	60.7 ± 5.4	56.4 ± 7.3	64.8 ± 3.8	70.1 ± 5.6	108030
Autoencoder	65.4 ± 6.6	69.3 ± 4.2	61.9 ± 5.2	60.8 ± 3.7	21600
Ours	71.8 ± 4.2	70.8 ± 3.8	65.9 ± 4.2	65.9 ± 4.8	1200