基于卷积辅助自注意力的胸部疾病分类网络

张自然,李锵,关欣

Classification network for chest disease based on convolution-assisted self-attention

Ziran ZHANG,Qiang LI,Xin GUAN

表 1 不同胸部疾病分类网络在ChestX-Ray14测试集上的结果比较

Tab.1 Comparison of result of different chest disease classification network on ChestX-Ray14 test set

疾病类别 AUC MXT TransDD PCAN PCSANet LCT CheXGAT SSGE ML-LGL CAWSNet 肺不张 0.798 0.791 0.785 0.807 0.789 0.787 0.792 0.782 0.829 心脏肿大 0.896 0.885 0.897 0.910 0.889 0.879 0.892 0.904 0.918 积液 0.842 0.842 0.837 0.879 0.842 0.837 0.840 0.835 0.892 浸润 0.719 0.715 0.706 0.698 0.694 0.699 0.714 0.707 0.726 肿块 0.856 0.837 0.834 0.824 0.843 0.839 0.848 0.853 0.857 结节 0.809 0.803 0.786 0.750 0.803 0.793 0.812 0.779 0.784 肺炎 0.758 0.745 0.730 0.750 0.742 0.741 0.733 0.739 0.782 气胸 0.879 0.885 0.871 0.850 0.896 0.879 0.885 0.889 0.903 肺实变 0.759 0.753 0.763 0.802 0.757 0.755 0.753 0.771 0.820 水肿 0.849 0.859 0.849 0.888 0.858 0.851 0.848 0.866 0.906 肺气肿 0.906 0.944 0.921 0.890 0.944 0.945 0.948 0.949 0.935 纤维化 0.847 0.849 0.817 0.812 0.863 0.842 0.827 0.846 0.827 胸腔积液 0.800 0.803 0.791 0.768 0.799 0.794 0.795 0.787 0.817 疝气 0.913 0.924 0.943 0.915 0.915 0.931 0.932 0.907 0.939 平均值 0.830 0.831 0.824 0.825 0.831 0.827 0.830 0.830 0.853