一、摘要

There is large consent that successful training of deep networks requires many thousand annotated training samples. In this paper, we present a network and training strategy that relies on the strong use of data augmentation to use the available annotated samples more effiffifficiently. The architecture consists of a contracting path to capture context and a symmetric expanding path that enables precise localization. We show that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks. Using the same network trained on transmitted light microscopy images (phase contrast and DIC) we won the ISBI cell tracking challenge 2015 in these categories by a large margin. Moreover, the network is fast. Segmentation of a 512x512 image takes less than a second on a recent GPU. The full implementation (based on Caffffe) and the trained networks are available at http://lmb.informatik.uni-freiburg.de/people/ronneber/u-net.

该体系结构包括一个可以捕获上下文的收缩路径和一个能够实现精确定位的对称扩展路径
只需要很少的训练数据（标注图像）就能达到非常低的错误率

二、背景

深度网络受限于可得到训练集的大小及数量，更大更深的网络以及更大的数据集可以更好的提升性能
医学上很难得到大量的数据
当时性能最好的网络：滑动窗口卷积网络（a sliding-window convolutional network）
- 优点1：能够进行定位
- 优点2：在图像块方面的训练数据远远大于训练图像的数量
- 缺点1：它相当慢，因为网络必须为每个patch单独运行，patches之间有大量的重叠则会出现重复训练
- 缺点2：定位精确度与context之间有trade-off。大的patch需要更大的pooilng层，使得定位误差增大；同时，小的patch使得网络提取出更少的语义信息

三、模型

Untitled

**收缩路径：**遵循卷积网络的典型架构。它包括重复应用两个3x3卷积（未填充卷积），每一个随后是一个校正线性单元(ReLU)和一个2x2最大池化操作，步幅2进行降采样。在每个降采样步骤中，我们将特征通道的数量增加一倍

**扩展路径：**它的每一步包括特征映射的上采样，即是2x2卷积（“上卷积”），即特征通道数量的一半，然后与收缩路径中相应裁剪的特征映射的连接，以及两个3x3卷积，然后是一个ReLU。由于在每次卷积中都丢失了边界像素，因此进行裁剪是必要的