Skip to main content
. 2023 Sep 1;11(9):2441. doi: 10.3390/biomedicines11092441

Table 2.

The classifier-wise results of the proposed ASNet.

No Generation Method AS Axial
Accuracy (%)
AS Coronal Accuracy (%) AS Contrast-
Enhanced Accuracy (%)
1 DenseNet201 fc1000 layer NCA kNN 98.81 94.16 92.61
2 fc1000 layer Chi2 kNN 97.51 93.07 90.91
3 fc1000 layer RF kNN 98.71 94.66 91.80
4 avg_pool layer NCA kNN 99.65 99.10 95.13
5 avg_pool layer Chi2 kNN 99.15 98.30 94.40
6 avg_pool layer RF kNN 99.60 98.90 95.05
7 ResNet50 fc1000 layer NCA kNN 98.41 94.66 93.10
8 fc1000 layer Chi2 kNN 97.11 92.12 90.83
9 fc1000 layer RF kNN 98.56 94.16 92.37
10 avg_pool layer NCA kNN 99.15 98.50 93.18
11 avg_pool layer Chi2 kNN 98.81 97.11 91.48
12 avg_pool layer RF kNN 99.10 98.00 93.34
13 ShuffleNet Node200 layer NCA kNN 98.76 91.37 92.45
14 Node200 layer Chi2 kNN 97.26 89.73 90.34
15 Node200 layer RF kNN 97.76 90.97 91.31
16 Node202 layer NCA kNN 98.71 95.61 92.78
17 Node202 layer Chi2 kNN 98.76 94.81 91.48
18 Node202 layer RF kNN 98.66 96.36 91.72