U-net:Convolutional Networks for Biomedical Image Segmentation

U-Net：Convolutional Networks for Biomedical Image Segmentation

Abstract ：

The architecture consists of a contracting path to capture context and a symmetric expanding path that enables precise localization

Introduction ：

• The convolutional network，their success was limited due to the size of the available training sets and the size of the considered networks.

• In many visual tasks，especially in biomedical image processing，the desired output should include localization，i.e.，a class label is supposed to be assigned to each pixel .

• Ciresan et al：trained a network in a sliding-window setup to predict the class label of each pixel by providing a local region (patch) around that pixel as input .

  • advantages：
    • this network can localize
    • the training data in terms of patches is much larger than the number of training images
  • disadvantages：
    • it is quite slow because the network must be run separately for each patch , and there is a lot of redundancy due to overlapping patches.
    • there is a trade-off between localization accuracy and the use of context.

• More recent approaches proposed a classifier output that takes into account the features from multiple layers.

  多尺度融合的深度网络，把某一个像素为中心的不同大小的patch作为多个通道输入到深度网络中学习

• In this paper , “fully convolutional network” , the main idea is to supplement a usual contracting network by successive layers , where pooling operators are replaced by upsampling operators .

• In the upsampling part we have alse a large number of feature channels , which allow the network to propagate context information to higher resolution layers . The segmentation map only contains the pixels , for which the full context is available in the input image . To predict the pixels in the border region of the image , the missing context is extrapolated by mirroring the input image .

This strategy allows the seamless segmentation of arbitrarily large images by an overlap-tile strategy.

​

• There is very little training data available , we use excessive data augmentation by applying elastic deformations to the available training images .
• Another challenge is the separation of touching objects of the same class . We propose the use of a weighted loss , where the separating background labels between touching cells obtain a larger weight in the loss function

Network Architecture

Training

• SGD

• unpadded

• larger input tiles over a large batch size

• softmax

• The separation border computed using morphological operations

  中间项链的部分像素，人为的提高权重，让network可以重点学习这些特征，以便于分割

• draw the initial weights from a Gaussian distribution

Data Augmentation

Data Augmentation is essential to teach the network the desired invariance and robustness properties , when only few training samples are available .

• Need shift and rotation invariance as well as robustness to deformations and gray value variations
• Generate smooth deformations using random displacement vectors on a coarse 3 by 3 grid .

Experiments

• 30 images(512*512 pixels)
• ground truth
• membranes
• The u-net (averaged over 7 rotated version of the input data)

Conclusion

​ The u-net architecture achieves very good performance on very different biomedical segmentation applications . Thanks to data augmentation with elastic deformations , it only needs very few annotated images and has a very reasonable training time .