Table 1 Comparison of networks containing regular convolution and dilated convolution based on different backbones.
From: Semantic segmentation of COVID-19 lesions with a multiscale dilated convolutional network
Backbone | Methods | Dice (%) | Sen. (%) | Spec. (%) | mIoU (%) |
|---|---|---|---|---|---|
VGG-16 | Baseline | 70.7 | 69.9 | 87.5 | 65.2 |
Dilated convolution | 72.6 | 70.2 | 87.4 | 67.5 | |
ResNet-50 | Baseline | 72.5 | 71.7 | 88.5 | 68.3 |
Dilated convolution | 74.2 | 75.1 | 91.1 | 70.8 |
