[论文解读] Domain Generalization for Medical Imaging Classification with Linear-Dependency Regularization

Haoliang Li, Yufei Wang+4 more|arXiv (Cornell University)|Sep 27, 2020

Domain Adaptation and Few-Shot Learning参考文献 45被引用 81

一句话总结

引入 Linear-Dependency Domain Generalization (LDDG)，通过强制潜在空间秩约束和分布对齐来提高跨域泛化，在没有目标域数据的情况下用于医学影像。

ABSTRACT

Recently, we have witnessed great progress in the field of medical imaging classification by adopting deep neural networks. However, the recent advanced models still require accessing sufficiently large and representative datasets for training, which is often unfeasible in clinically realistic environments. When trained on limited datasets, the deep neural network is lack of generalization capability, as the trained deep neural network on data within a certain distribution (e.g. the data captured by a certain device vendor or patient population) may not be able to generalize to the data with another distribution. In this paper, we introduce a simple but effective approach to improve the generalization capability of deep neural networks in the field of medical imaging classification. Motivated by the observation that the domain variability of the medical images is to some extent compact, we propose to learn a representative feature space through variational encoding with a novel linear-dependency regularization term to capture the shareable information among medical data collected from different domains. As a result, the trained neural network is expected to equip with better generalization capability to the "unseen" medical data. Experimental results on two challenging medical imaging classification tasks indicate that our method can achieve better cross-domain generalization capability compared with state-of-the-art baselines.

研究动机与目标

• 在目标域数据不可用时，激励并解决医学影像中的域迁移问题。
• 学习一个能够跨域共享信息的代表性潜在特征空间。
• 对潜在特征进行秩的正则化，使秩为类别数目（固定秩）。
• 将潜在特征分布对齐到高斯先验以降低对源域的过拟合。

提出的方法

• 模型使用特征提取器 Q_theta、变分编码器 F_omega 和线性分类器 T_phi。
• 对潜在特征施加基于秩的正则化，使秩(Z)=C（类别数）以捕获类别信息。
• 基于 KL 散度的分布对齐，将来自多个源域的潜在特征映射到高斯先验 N(0,1)。
• 以复合损失进行训练：L_obj = sum L_c(y_hat, y) + lambda1 L_rank + lambda2 KL(q(Z|X) || N(0,1))。
• 在假设 1-2 下给出对目标域风险的理论上界，并证明相关定理。
• 在皮肤病变分类和脊髓灰质分割上进行实验，以证明跨域泛化的提升。

实验结果

研究问题

• RQ1 潜在空间秩约束是否能够捕获跨多个医学影像域的共享信息，从而提高对未见域的泛化能力？
• RQ2 通过 KL 散度将潜在特征对齐到预定义先验，是否能减少对源域的过拟合并提高目标域性能？
• RQ3 提出的 LDDG 相对于皮肤病变分类和脊髓分割任务中的最先进域泛化基线的比较？
• RQ4 改变潜在空间秩对泛化性能有何影响？

主要发现

• LDDG 在跨域皮肤病变分类任务中超过基线（DeepAll、MASF、MLDG、CCSA），覆盖多个目标域。
• 在表1中的目标域平均结果上，LDDG 在所比较方法中表现最佳。
• 消融实验显示秩和 KL 两个分量都能提升泛化，其中秩设为 C（等于类别数）在许多设置中达到峰值精度。
• 使用核范数的秩正则化相对于提出的秩(C)约束效果较差，凸显了将类别信息纳入的重要性。
• 在消融实验中，结合秩和 KL 的效果优于任一单独成分。