School of Information Engineering, Zhengzhou University, Zhengzhou, 450001, China.
Collaborative Innovation Center for Internet Healthcare, Zhengzhou University, Zhengzhou, 450052, China.
Med Phys. 2021 Mar;48(3):1157-1167. doi: 10.1002/mp.14671. Epub 2021 Jan 10.
Breast mass segmentation is a prerequisite step in the use of computer-aided tools designed for breast cancer diagnosis and treatment planning. However, mass segmentation remains challenging due to the low contrast, irregular shapes, and fuzzy boundaries of masses. In this work, we propose a mammography mass segmentation model for improving segmentation performance.
We propose a mammography mass segmentation model called SAP-cGAN, which is based on an improved conditional generative adversarial network (cGAN). We introduce a superpixel average pooling layer into the cGAN decoder, which utilizes superpixels as a pooling layout to improve boundary segmentation. In addition, we adopt a multiscale input strategy to enable the network to learn scale-invariant features with increased robustness. The performance of the model is evaluated with two public datasets: CBIS-DDSM and INbreast. Moreover, ablation analysis is conducted to evaluate further the individual contribution of each block to the performance of the network.
Dice and Jaccard scores of 93.37% and 87.57%, respectively, are obtained for the CBIS-DDSM dataset. The Dice and Jaccard scores for the INbreast dataset are 91.54% and 84.40%, respectively. These results indicate that our proposed model outperforms current state-of-the-art breast mass segmentation methods. The superpixel average pooling layer and multiscale input strategy has improved the Dice and Jaccard scores of the original cGAN by 7.8% and 12.79%, respectively.
Adversarial learning with the addition of a superpixel average pooling layer and multiscale input strategy can encourage the Generator network to generate masks with increased realism and improve breast mass segmentation performance through the minimax game between the Generator network and Discriminator network.
乳腺肿块分割是计算机辅助工具用于乳腺癌诊断和治疗计划的前提步骤。然而，由于肿块的对比度低、形状不规则和边界模糊，肿块分割仍然具有挑战性。在这项工作中，我们提出了一种用于提高分割性能的乳腺钼靶肿块分割模型。
我们提出了一种称为 SAP-cGAN 的乳腺钼靶肿块分割模型，该模型基于改进的条件生成对抗网络（cGAN）。我们在 cGAN 解码器中引入了超像素平均池化层，该层利用超像素作为池化布局来提高边界分割。此外，我们采用多尺度输入策略，使网络能够学习具有更高鲁棒性的尺度不变特征。使用两个公共数据集 CBIS-DDSM 和 INbreast 来评估模型的性能。此外，还进行了消融分析，以进一步评估每个模块对网络性能的贡献。
在 CBIS-DDSM 数据集上，获得了 93.37%的 Dice 分数和 87.57%的 Jaccard 分数。在 INbreast 数据集上，Dice 分数和 Jaccard 分数分别为 91.54%和 84.40%。这些结果表明，我们提出的模型优于当前最先进的乳腺肿块分割方法。超像素平均池化层和多尺度输入策略分别将原始 cGAN 的 Dice 分数和 Jaccard 分数提高了 7.8%和 12.79%。
在对抗学习中加入超像素平均池化层和多尺度输入策略，可以通过生成器网络和判别器网络之间的最小极大博弈，鼓励生成器网络生成更真实的掩模，从而提高乳腺肿块分割性能。
