Table 4 Comparison of image-level detection performance across different models.
From: DLA-Net: dual lesion attention network for classification of pneumoconiosis using chest X-ray images
Models | Binary classification | ||||
|---|---|---|---|---|---|
ACC | SEN | SPE | F1 | AUC | |
DenseNet121 | 0.8562 ± 2.75 | 0.8562 ± 2.75 | 0.8558 ± 2.77 | 0.8555 ± 2.85 | 0.7958 ± 1.57 |
ResNet50 | 0.8459 ± 2.34 | 0.8459 ± 2.34 | 0.8455 ± 2.32 | 0.8457 ± 2.34 | 0.7947 ± 2.95 |
Inceptionv3 | 0.8538 ± 1.04 | 0.8538 ± 1.04 | 0.8535 ± 1.02 | 0.8536 ± 1.03 | 0.7981 ± 0.91 |
Xception | 0.8563 ± 3.43 | 0.8563 ± 3.43 | 0.8557 ± 3.46 | 0.8551 ± 3.54 | 0.8013 ± 2.51 |
EfficientNetB4 | 0.8570 ± 1.16 | 0.8570 ± 1.16 | 0.8564 ± 1.16 | 0.8561 ± 1.21 | 0.8071 ± 1.18 |
DLA-Net (Proposed) | 0.8841 ± 2.22 | 0.8841 ± 2.22 | 0.8840 ± 2.22 | 0.8841 ± 2.22 | 0.8291 ± 2.29 |
