本文整理了最近看的论文，主要方向是脑肿瘤分割，由论文链接，论文出发点，以及论文的创新点构成。
Densely Connected Convolutional Networks
Standpoint
Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output.
Innovation
Design densely connected convolutional networks with shorter connections.
DRINet for Medical Image Segmenation
Standpoint
These convolution layers learn representative features of input images and construct segmentation based on the features. However, the features learned by standard convolution layers are not distinctive when the differences among different categoriesare subtle in terms of intensity, location,shape, and size.
Innovation
A novel combination of the dense connections with the inception structure to address segmentation problems. The use of dense connection blocks, residual inception blocks, and the unpooling blocks achieve high performance while maintaining computational efficiency;
Autofocus Layer for Semantic Segmentation
Standpoint
For high performance, segmentation algorithms are required to use multi-scale context [6], while still aiming for pixel-level accuracy. Multi-scale processing provides detailed cues, such as texture information of a structure, combined with contextual information, such as a structure’s surroundings, which can facilitate decisions that are ambiguous when based only on local context.
Innovation
They propose the autofocus convolutional layer for semantic segmentation with the objective of enhancing the capabilities of neural networks for multi-scale processing. Autofocus layers adaptively change the size of the e_ective receptive field based on the processed context to generate more powerful features. This is achieved by parallelising mul-tiple convolutional layers with di_erent dilation rates, combined by an attention mechanism that learns to focus on the optimal scales driven by context.
Efficient Brain Tumor Segmentation with Multiscale Two-Pathway-Group Conventional Neural Networks
Standpoint
Manual segmentation of the brain tumors for cancerdiagnosis from MRI images is a difficult, tedious and timeconsuming task. The accuracy and the robustness of brain tumor segmentation, therefore, are crucial for the diagnosis, treatment planning, and treatment outcome evaluation.
Traditional methods of Deep learning such as Convolutional Neural Networks require a large amount of annotated data to learn from, which is often difficult to obtain in the medical domain.
Innovation
They describe a new model Two-Pathway-Group CNN architecture for brain tumor segmentation, which exploits local features and global contextual features simultaneously.
Efficient Multi-scale 3D CNN with Fully Connected CRF for Accurate Brain lesion Segmentation
Standpoint
The heterogeneous appearance of lesions including the large variability in location, size, shape and frequency make it difficult to devise effective segmentation rules. It is thus highly non-trivial to delineate contu- sions, oedema and haemorrhages in TBI, or sub-components of brain tumours such as proliferating cells and necrotic core.
Innovation
They propose a dual pathway, 11-layers deep, three-dimensional Convolutional Neural Network for the challenging task of brain lesion segmentation.
HyperDense-Net: A Hyper-densely Connected CNN for Multi-modal Image Segmentation
Standpoint
Dense connections have attracted substantial attention in computer vision because they facilitate gradient flow and implicit deep supervision during training. Particularly, DenseNet, which connects each layer to every other layer in a feed-forward fashion, has shown impressive performances in natural image classification tasks.
Innovation
They propose HyperDenseNet, a 3D fully convolutional neural network that extends the definition of dense connectivity to multi-modal segmentation problems. Each imaging modality has a path, and dense connections occur not only between the pairs of layers within the same path, but also between those across different paths.
A Deep Learning model integrating FCNNs and CRFs for brain tumor segmentation
Standpoint
Accurate and reliable brain tumor segmentation is a critical component in cancer diagnosis, treatment planning, and treatment outcome evaluation.
Innovation
Build upon successful deep learning techniques, a novel brain tumor segmentation method is developed by integrating fully convolutional neural networks (FCNNs) and Conditional Random Fields (CRFs) in a unified framework to obtain segmentation results with appearance and spatial consistency.
U-Net: Convolutional Networks for Biomedical Image Segmentation
Standpoint
There is large consent that successful training of deep networks requires many thousand annotated training samples.
Innovation
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 efficiently.
Review
This architecture is very useful in many medical image segmentation task, so is it the best architecture?
The novel architecture is not emphasized in this paper, but this paper propose U-Net that has been a popular network architecture.
