Department of Radiation Oncology, Medical University of Vienna, Vienna, Austria; Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria.
Technical University of Vienna, Vienna, Austria.
Z Med Phys. 2022 Aug;32(3):361-368. doi: 10.1016/j.zemedi.2021.11.006. Epub 2021 Dec 18.
For image translational tasks, the application of deep learning methods showed that Generative Adversarial Network (GAN) architectures outperform the traditional U-Net networks, when using the same training data size. This study investigates whether this performance boost can also be expected for segmentation tasks with small training dataset size.
MATERIALS/METHODS: Two models were trained on varying training dataset sizes ranging from 1-100 patients: a) U-Net and b) U-Net with patch discriminator (conditional GAN). The performance of both models to segment the male pelvis on CT-data was evaluated (Dice similarity coefficient, Hausdorff) with respect to training data size.
No significant differences were observed between the U-Net and cGAN when the models were trained with the same training sizes up to 100 patients. The training dataset size had a significant impact on the models' performances, with vast improvements when increasing dataset sizes from 1 to 20 patients.
When introducing GANs for the segmentation task no significant performance boost was observed in our experiments, even in segmentation models developed on small datasets.
在图像转换任务中，与传统的 U-Net 网络相比，当使用相同的训练数据大小时，生成对抗网络 (GAN) 架构在应用深度学习方法时表现更好。本研究调查了在小训练数据集大小的分割任务中是否也可以期待这种性能提升。
材料/方法：在从 1 到 100 名患者的不同训练数据集大小上训练了两个模型：a) U-Net 和 b) 具有补丁判别器（条件 GAN）的 U-Net。评估了两种模型在 CT 数据上分割男性骨盆的性能（Dice 相似系数、Hausdorff），并考虑了训练数据大小。
当使用相同的训练大小（最多 100 名患者）训练模型时，U-Net 和 cGAN 之间没有观察到显著差异。训练数据集大小对模型性能有显著影响，当从 1 名患者增加到 20 名患者时，性能显著提高。
在引入 GAN 进行分割任务时，即使在小型数据集上开发的分割模型中，我们的实验也没有观察到明显的性能提升。
