Biomedical Engineering and Physics, Amsterdam UMC, Location AMC, Meibergdreef 15, 1105 AZ Amsterdam, the Netherlands; University of Amsterdam, Science Park 904, 1098 XH Amsterdam, the Netherlands.
University of Amsterdam, Science Park 904, 1098 XH Amsterdam, the Netherlands.
Comput Methods Programs Biomed. 2022 Feb;214:106539. doi: 10.1016/j.cmpb.2021.106539. Epub 2021 Nov 23.
Transfer learning is a valuable approach to perform medical image segmentation in settings with limited cases available for training convolutional neural networks (CNN). Both the source task and the source domain influence transfer learning performance on a given target medical image segmentation task. This study aims to assess transfer learning-based medical segmentation task performance for various source task and domain combinations.
CNNs were pre-trained on classification, segmentation, and self-supervised tasks on two domains: natural images and T1 brain MRI. Next, these CNNs were fine-tuned on three target T1 brain MRI segmentation tasks: stroke lesion, MS lesions, and brain anatomy segmentation. In all experiments, the CNN architecture and transfer learning strategy were the same. The segmentation accuracy on all target tasks was evaluated using the mIOU or Dice coefficients. The detection accuracy was evaluated for the stroke and MS lesion target tasks only.
CNNs pre-trained on a segmentation task on the same domain as the target tasks resulted in higher or similar segmentation accuracy compared to other source task and domain combinations. Pre-training a CNN on ImageNet resulted in a comparable, but not consistently higher lesion detection rate, despite the amount of training data used being 10 times larger.
This study suggests that optimal transfer learning for medical segmentation is achieved with a similar task and domain for pre-training. As a result, CNNs can be effectively pre-trained on smaller datasets by selecting a source domain and task similar to the target domain and task.
在用于训练卷积神经网络（CNN）的病例有限的情况下，迁移学习是一种进行医学图像分割的有效方法。源任务和源域都会影响给定的目标医学图像分割任务中迁移学习的性能。本研究旨在评估各种源任务和源域组合的基于迁移学习的医学分割任务性能。
在两个域（自然图像和 T1 脑 MRI）上，对分类、分割和自监督任务进行了 CNN 预训练。接下来，在三个目标 T1 脑 MRI 分割任务（中风病变、MS 病变和脑解剖分割）上对这些 CNN 进行微调。在所有实验中，CNN 架构和迁移学习策略都是相同的。使用 mIOU 或 Dice 系数评估所有目标任务的分割准确性。仅对中风和 MS 病变目标任务评估检测准确性。
与其他源任务和源域组合相比，在与目标任务相同的域上进行预训练的分割任务的 CNN 导致更高或相似的分割准确性。尽管使用的训练数据量是 10 倍，但在 ImageNet 上预训练 CNN 导致的病变检测率相当，但并非始终更高。
本研究表明，对于医学分割，最佳的迁移学习是通过与目标域和任务相似的源域和任务进行预训练来实现的。因此，可以通过选择与目标域和任务相似的源域和任务，有效地在较小的数据集上对 CNN 进行预训练。
