CE-Net：深度学习医疗图像分割的创新上下文编码器

115 浏览量更新于2024-08-28 收藏 2.99MB PDF 举报

本文档主要介绍了一篇名为"CE-Net：用于2D医学图像分割的上下文编码器网络"的研究论文，发表在《IEEE Transactions on Medical Imaging》的第38卷第10期，2019年10月刊。医学图像分割是医疗图像分析中的关键步骤，随着卷积神经网络（CNN）在图像处理领域的快速发展，深度学习已经在诸如视网膜盘分割、血管检测、肺部分割和细胞分割等任务中得到广泛应用。然而，传统的U-Net方法由于连续池化和步进卷积操作可能导致一些空间信息的丢失。作者Zaiwang Gu、Jun Cheng、Huazhu Fu、Kang Zhou、Huaying Hao、Yitian Zhao、Tianyang Zhang、Shenghua Gao和Jiang Liu针对这一问题，提出了CE-Net，一个旨在捕捉更高级别信息并保留空间信息的上下文编码网络。CE-Net主要由三个核心组件构成： 1. 特征编码模块（Feature Encoder Module）：这个部分使用预训练的ResNet块作为固定特征提取器，ResNet的深度学习结构能够提取出丰富的多层次特征，为后续处理提供高质量的基础。 2. 上下文提取器（Context Extractor）：这是CE-Net的关键创新点，通过设计有效的架构，该模块能够增强对图像上下文的理解，避免了传统方法中信息的丢失，有助于保持图像的空间连贯性。 3. 特征解码模块（Feature Decoder Module）：此模块负责将编码后的特征逐步解码回原始尺寸，同时结合上采样技术来恢复被压缩的空间细节，与编码过程形成互补，以实现更精确的分割结果。 CE-Net通过巧妙地结合上下文信息和空间分辨率的保持，改进了现有医学图像分割的性能，为2D医学图像的自动化分析提供了新的解决方案。这种方法不仅适用于眼科、血管和肺部等领域，还可能扩展到其他需要精细分割的应用场景，具有较高的实用价值和研究意义。

et al.

: CE-NET FOR 2D MEDICAL IMAGE SEGMENTATION 2283

Fig. 1. Illustration of the proposed CE-Net. Firstly, the images are fed into a feature encoder module, where the ResNet-34 block pretrained from

ImageNet is used to replace the original U-Net encoder block. The context extractor is proposed to generate more high-level semantic feature maps.

It contains a dense atrous convolution (DAC) block and a residual multi-kernel pooling (RMP) block. Finally, the extracted features are fed into the

feature decoder module. In this paper, we adopt a decoder block to enlarge the feature size, replacing the original up-sampling operation. T he

decoder block contains 1×1 convolution and 3×3 deconvolution operations. Based on skip connection and the decoder block, we obtain the mask

as the segmentation prediction map.

from spatial pyramid pooling [55]. The RMP block fur-

ther encodes the multi-scale context features of the object

extracted from the DAC module by employing various size

pooling operations, without the extra learning weights. In

summary, the DAC block is proposed to extract enriched

feature representations with multi-scale atrous convolutions,

followed by the RMP block for further context information

with multi-scale pooling operations. Integrating the newly

proposed DAC block and the RMP block with the backbone

encoder-decoder structure, we propose a novel context encoder

network named as CE-Net. It relies on the DAC block and the

RMP block to get more abstract features and preserve more

spatial information to boost the performance of medical image

segmentation.

The main contributions of this work are summarized as

follows:

1) We propose a DAC block and RMP block to capture

more high-level features and preserve more spatial infor-

mation.

2) We integrate the proposed DAC block and RMP block

with encoder-decoder structure for medical image seg-

mentation.

3) We apply the proposed method in different tasks includ-

ing optic disc segmentation, retinal vessel detection,

lung segmentation, cell contour segmentation and retinal

OCT layer segmentation. Results show that the proposed

method outperforms the state-of-the-art methods in these

different tasks.

The remainder of this paper is organized as follows.

Section II introduces the proposed method in details.

Section III presents the experimental results and discussions.

In Section IV, we draw some conclusions.

II. M

ETHOD

The proposed CE-Net consists of three major parts: the

feature encoder module, the context extractor module, and the

feature decoder module, as shown in Fig. 1.

A. Feature Encoder Module

In U-Net architecture, each block of encoder contains two

convolution layers and one max pooling layer. In the proposed

method, we replace it with the pretrained ResNet-34 [53]

in the feature encoder module, which retains the ﬁrst four

feature extracting blocks without the average pooling layer

and the fully connected layers. Compared with the original

block, ResNet adds shortcut mechanism to avoid the gradient

vanishing and accelerate the network convergence, as shown

in Fig. 1(b). For convenience, we use the modiﬁed U-net with

pretrained ResNet as backbone approach.

B. Conte xt Extractor Module

The context extractor module is a newly proposed module,

consisting of the DAC block and the RMP block. This module

extracts context semantic information and generates more

high-level feature maps.

剩余11页未读，继续阅读

weixin_38697557

粉丝: 8
资源: 921

CE-Net：深度学习医疗图像分割的创新上下文编码器

CE模块源代码分析和运行原理

Python-CENet用于2D医学图像分割的上下文编码器网络

Python-CGNet一种用于语义分割的轻量级上下文引导网络

LSTM情感分析在多模态数据：从文本到图像，全面解读情感

图像分割中的深度学习方法综述

高级图像处理：利用神经网络在SimpleCV中实现复杂处理

语义分割 vs 实例分割：区别与联系

Python Pango调试秘籍：如何快速定位和解决渲染问题

深度学习性能优化：掌握模型调优的5个秘诀

PyQt4.QtGui绘图进阶：掌握自定义控件和绘图技巧

最新资源