Table 2 Comparison of classification performance of different models at zone level.
From: DLA-Net: dual lesion attention network for classification of pneumoconiosis using chest X-ray images
Zone | DenseNet121 | Xception | EfficientB4 | Proposed | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Acc | F1 | AUC | Acc | F1 | AUC | Acc | F1 | AUC | Acc | F1 | AUC | |
LLZ | 0.826 ± 0.96 | 0.738 ± 0.66 | 0.847 ± 1.40 | 0.821 ± 1.20 | 0.744 ± 1.49 | 0.843 ± 1.92 | 0.826 ± 1.16 | 0.735 ± 0.83 | 0.851 ± 1.37 | 0.827 ± 1.11 | 0.739 ± 1.66 | 0.844 ± 2.44 |
LMZ | 0.867 ± 2.11 | 0.773 ± 3.51 | 0.893 ± 1.91 | 0.867 ± 1.95 | 0.777 ± 2.62 | 0.879 ± 2.67 | 0.858 ± 1.76 | 0.772 ± 2.59 | 0.875 ± 1.78 | 0.867 ± 1.59 | 0.780 ± 2.07 | 0.896 ± 2.12 |
LUZ | 0.888 ± 1.22 | 0.819 ± 2.16 | 0.914 ± 1.08 | 0.885 ± 1.71 | 0.811 ± 2.87 | 0.901 ± 2.09 | 0.891 ± 1.98 | 0.820 ± 2.82 | 0.909 ± 1.32 | 0.889 ± 1.51 | 0.816 ± 2.17 | 0.898 ± 1.81 |
RLZ | 0.824 ± 2.64 | 0.727 ± 3.17 | 0.868 ± 2.79 | 0.825 ± 2.72 | 0.733 ± 3.43 | 0.873 ± 2.41 | 0.818 ± 1.03 | 0.728 ± 1.69 | 0.862 ± 1.70 | 0.841 ± 1.83 | 0.740 ± 2.90 | 0.885 ± 1.92 |
RMZ | 0.844 ± 2.24 | 0.743 ± 2.51 | 0.887 ± 1.91 | 0.841 ± 1.50 | 0.754 ± 3.15 | 0.893 ± 2.17 | 0.853 ± 1.03 | 0.762 ± 1.97 | 0.901 ± 0.93 | 0.856 ± 0.82 | 0.777 ± 2.29 | 0.901 ± 1.66 |
RUZ | 0.885 ± 0.87 | 0.807 ± 2.15 | 0.911 ± 1.04 | 0.891 ± 0.99 | 0.806 ± 1.27 | 0.908 ± 0.98 | 0.888 ± 0.91 | 0.816 ± 1.56 | 0.912 ± 2.16 | 0.888 ± 0.75 | 0.811 ± 1.48 | 0.914 ± 1.38 |
Avg. | 0.856 ± 1.67 | 0.768 ± 2.36 | 0.886 ± 1.68 | 0.8555 ± 1.67 | 0.771 ± 2.47 | 0.883 ± 2.04 | 0.856 ± 1.31 | 0.772 ± 1.91 | 0.885 ± 1.54 | 0.861 ± 1.48 | 0.777 ± 2.07 | 0.890 ± 1.53 |
