Currently, breast tissue images are primarily classified by pathologists, which is time-consuming and subjective. Deep learning, however, can perform this task with the utmost precision. In order to achieve an improved performance, a large number of annotated datasets are required to train the network, which is a challenging task in the medical field. In this paper, we propose an intelligent system, based on generative adversarial networks (GANs) and a convolution neural network (CNN) for the automatic classification of breast cancer, using optical coherence tomography (OCT) images. In this network, the GAN is used to generate synthetic datasets and to further utilize these synthetic datasets to increase the quantity of information, so as to improve the classification performance of the CNN. Our method is demonstrated by means of a limited set of OCT images of breast tissue. The classification performance of our method, using only the classic data increase, yielded a sensitivity...

中文翻译：

---

基于光学相干断层图像的基于生成对抗网络-卷积神经网络的乳腺癌分类

当前，乳房组织图像主要由病理学家分类，这既费时又主观。但是，深度学习可以最高精度地执行此任务。为了获得改进的性能，需要大量带注释的数据集来训练网络，这在医学领域是一项艰巨的任务。在本文中，我们提出了一个基于生成对抗网络（GAN）和卷积神经网络（CNN）的智能系统，用于使用光学相干断层扫描（OCT）图像对乳腺癌进行自动分类。在该网络中，GAN用于生成综合数据集，并进一步利用这些综合数据集来增加信息量，从而提高CNN的分类性能。我们的方法通过一组有限的乳腺组织OCT图像得到证明。我们方法的分类性能，仅使用经典数据增加，就产生了灵敏度。