论文标题
常规成像中的自动肺部分割主要是数据多样性问题，而不是方法论问题
Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem
论文作者
论文摘要
解剖结构的自动分割是图像分析的关键步骤。对于计算机断层扫描中的肺部分割，存在各种方法，涉及在不同数据集上训练和验证的复杂管道。但是，跨疾病的这些方法的临床适用性仍然有限。我们比较了在各种数据集中训练的四种通用深度学习方法和两种随时可用的肺部分割算法。我们对常规成像数据进行了评估，该数据具有六个以上不同的疾病模式和三个已发表的数据集。使用不同的深度学习方法，测试数据集上的平均骰子相似性系数（DSC）的变化不超过0.02。与在公共数据集中的培训相比，接受标准方法（U-NET）进行多样化的常规数据集（n = 36）时，一种标准方法（U-NET）会产生更高的DSC（0.97 $ \ pm $ 0.05）。与参考方法相比，对涵盖多种疾病的常规数据（n = 231）进行了训练，U-NET的DSC为0.98 $ \ pm $ 0.03，而0.94 $ \ pm $ \ pm $ 0.12（p = 0.024）。
Automated segmentation of anatomical structures is a crucial step in image analysis. For lung segmentation in computed tomography, a variety of approaches exist, involving sophisticated pipelines trained and validated on different datasets. However, the clinical applicability of these approaches across diseases remains limited. We compared four generic deep learning approaches trained on various datasets and two readily available lung segmentation algorithms. We performed evaluation on routine imaging data with more than six different disease patterns and three published data sets. Using different deep learning approaches, mean Dice similarity coefficients (DSCs) on test datasets varied not over 0.02. When trained on a diverse routine dataset (n = 36) a standard approach (U-net) yields a higher DSC (0.97 $\pm$ 0.05) compared to training on public datasets such as Lung Tissue Research Consortium (0.94 $\pm$ 0.13, p = 0.024) or Anatomy 3 (0.92 $\pm$ 0.15, p = 0.001). Trained on routine data (n = 231) covering multiple diseases, U-net compared to reference methods yields a DSC of 0.98 $\pm$ 0.03 versus 0.94 $\pm$ 0.12 (p = 0.024).
