CN113744210A

CN113744210A - Heart segmentation method based on multi-scale attention U-net network

Info

Publication number: CN113744210A
Application number: CN202110964471.4A
Authority: CN
Inventors: 崔晓娟; 白鑫昊; 杨铁军; 李磊; 樊超; 巩跃洪; 苗建雨
Original assignee: Henan University of Technology
Current assignee: Henan University of Technology
Priority date: 2021-08-22
Filing date: 2021-08-22
Publication date: 2021-12-03

Abstract

针对经典U形分割网络在分割心脏子结构精度不高的问题，本发明提出了一种多尺度注意力U‑Net网络用于心脏分割。在经典U‑Net结构基础上，算法首先将CT心脏图像进行裁剪预处理，通过减少背景像素的输入，接着采用Z‑score标准化方法进行归一化处理，以消除像素灰度分布范围的差异；然后引入注意力机制，充分利用浅层信息。在骨干网络中加入空间注意机制，在跳转连接中加入通道注意机制，使网络能够充分利用浅卷积层提取信息，保留有用信息，去除冗余信息；同时引入不同尺度卷积核的Inception模块，同时提取并融合不同尺度的特征信息，实现精确分割。本发明在心脏自动分割中具有良好的应用前景。Aiming at the problem that the classical U-shaped segmentation network has low accuracy in segmenting the substructure of the heart, the present invention proposes a multi-scale attention U-Net network for cardiac segmentation. Based on the classic U-Net structure, the algorithm firstly preprocesses the CT heart image, reduces the input of background pixels, and then uses the Z-score normalization method for normalization to eliminate the difference in the distribution range of pixel gray levels; Then an attention mechanism is introduced to make full use of shallow information. The spatial attention mechanism is added to the backbone network, and the channel attention mechanism is added to the jump connection, so that the network can make full use of the shallow convolution layer to extract information, retain useful information, and remove redundant information; at the same time, Inception modules with different scale convolution kernels are introduced. , while extracting and fusing feature information of different scales to achieve accurate segmentation. The invention has a good application prospect in the automatic segmentation of the heart.

Description

Heart segmentation method based on multi-scale attention U-net network

Technical Field

The invention belongs to the technical field of medical image processing, and particularly relates to a method for cardiac substructure segmentation.

Background

According to the heart disease and stroke statistical report of the American Heart Association (AHA) in 2019, it is pointed out that about 1055000 cases are expected to develop coronary heart disease in 2019 in the united states, including 720000 new and 335000 recurrent coronary artery cases, and this figure is also increasing year by year. At present, cardiovascular diseases are one of the main causes of human non-accidental death, the morbidity is high, and no fixed morbidity rule exists. Now, cardiovascular diseases gradually develop to young people and threaten the healthy life of human beings all the time. Accurate calculation, modeling and analysis of the entire cardiac structure is critical for research and application in the medical field for effective treatment and prevention of these diseases. At present, the cardiac substructure is generally segmented manually by doctors or experts according to the existing medical knowledge, medical conditions and clinical knowledge, the method is time-consuming and labor-consuming and has strong subjectivity, and the segmentation result is different from person to person. With the advent of large-scale labeled data and the development of computers, the realization of automatic segmentation of the heart using deep learning algorithms has become a hot spot of current research. Among them, Olaf Ronneberger proposed a U-Net model for medical image segmentation based on the Full Convolutional Network (FCN) model in 2015. Both U-Net and FCN have classical encoding-decoding topologies, but U-Net has a symmetric network structure and hopping connections, and the results of U-Net are superior to FCN in segmentation of cardiac images. Aiming at the problem that the heart image segmentation precision is difficult to improve, the improvement research performed by researchers on the basis of U-Net can be roughly divided into two types: improvement study based on 2D U-Net framework, improvement study based on 3D U-Net framework. Although the 2D network partitioning is less computationally and storage demanding, the 2D network partitioning typically discards spatial information between slices. 3D network segmentation can maximize the information between slice sequences. In summary, aiming at improving the segmentation accuracy, the invention provides a 3D U heart segmentation network AU-Net with attention mechanism, which is mainly used for solving the problem of low segmentation accuracy of the segmented heart substructure.

Disclosure of Invention

The invention aims to solve the problem of low precision of heart substructure segmentation, and provides an automatic method for accurately segmenting the heart substructure.

The invention is realized by the following technical scheme: a heart segmentation method based on a multi-scale attention U-net network. Firstly, cutting and scaling preprocessing are carried out on an image so as to reduce training parameters and cover global information; then 30000 times of training is carried out on the AMU-Net network of the invention, and the trained parameters are stored; and finally, segmenting the test data set by using the trained weight to finally obtain a segmentation result graph.

(1) Data preprocessing: image preprocessing first re-encodes the label data of the CT images of 10 training volume data in the MM-WHS2017 dataset to make them suitable for multi-classification tasks. The data is then randomly cropped, the image is cropped to 256 x 16 size, and light data enhancement techniques are applied to the data and label. On a random basis, the data is rotated between-15 and +15 degrees and scaled in between by 0.9-1.1. This ensures slight robustness and variability of network training.

(2) A training stage: in the training stage, the AMU-Net network provided by the invention is subjected to parameter training. And (3) learning model parameters of the training set by using a Tensorflow deep learning framework, setting the batch processing size to be 4 and setting the training iteration number to be 30000 by adopting an Adam optimizer. Selecting 50% of data in the training set as a verification set, training the data by using a cross entropy loss function, and selecting a weight coefficient when the verification set is minimally lost as a final training weight in a training stage in an experiment to provide a weight parameter for a subsequent testing stage.

The main innovation of the AU-Net network provided by the invention is as follows:

a. introducing a multi-scale inclusion module: by introducing the inclusion modules of convolution kernels with different scales, as shown in fig. 1, feature information with different scales is extracted and fused, global information and local information are better combined, and more features are obtained.

b. An attention mechanism is introduced: by introducing an attention mechanism, non-information areas are filtered, areas containing useful information are emphasized, the resolution of characteristic images is enhanced, the expression capacity of detail characteristics is improved, and more detail characteristics can be obtained.

(3) And (3) a testing stage: randomly selecting 10% of training volume data in the MM-WHS2017 data set as a test set. Firstly, a CT image with the size of 256 multiplied by 16 is directly input into a testing stage, and the final weight parameters obtained in a training stage are used for carrying out heart substructure segmentation on the testing image, so that a segmented heart substructure result graph is finally obtained.

The invention provides an improved AMU-Net heart image segmentation algorithm aiming at the defects that the heart image segmentation precision of a U-shaped convolution network is not high and the boundary of each region is fuzzy. By introducing a multi-scale inclusion module, feature information of different scales is extracted and fused. And an attention mechanism is introduced, the reusability of the features is increased, the shallow features are fused with the corresponding high-level features, and finally a trainable end-to-end segmentation algorithm is formed. Compared with the classic U-Net segmentation algorithm, the algorithm has a finer structure, can effectively solve the problems of over-segmentation and under-segmentation of heart segmentation, and has higher accuracy of segmentation results.

Drawings

Fig. 1 is a schematic diagram of a multi-scale inclusion structure.

Fig. 2 is a flow chart of a network training method for cardiac segmentation.

Detailed Description

To verify the cardiac substructure segmentation performance of the present invention, we selected the MM-WHS2017 dataset for training and testing.

The method comprises the steps of preprocessing CT image data, using Spyder software, and performing image normalization processing by image rotation, translation transformation and contrast enhancement.

Step two, training the AMU-Net network in Spyder software, wherein the batch _ size is 4, the learning _ rate is 0.001, adopting an Adama optimizer, using L2 regularization to prevent overfitting, and setting the regularization coefficient to be 0.0005. 60000 epochs are trained, the training set and the verification set are divided by the ratio of 1:1, the two processes are discontinuously trained, network parameters are adjusted until the network converges, and the training is finished.

And step three, testing the AMU-Net network by adopting a test set of the MM-WHS2017 data set. To evaluate the segmentation results, 2 common evaluation criteria, similarity coefficient (Dice) and Jaccard index, were used, as in table 1.

TABLE 1 comparison with other people's work (Dice index)

Experimental results show that the algorithm structure of the invention is finer, the problems of over-segmentation and under-segmentation of heart segmentation can be effectively solved, the segmentation result precision is higher, and the completeness and the accuracy of the heart segmentation can be ensured.

Claims

1. The invention discloses a heart segmentation method based on a multi-scale attention U-net network, which comprises the following steps of:

(1) data preprocessing: firstly, carrying out label data recoding on CT images of 10 training volume data in an MM-WHS2017 data set to enable the CT images to be suitable for a multi-classification task; then randomly cutting the data, cutting the image into 256 multiplied by 16 size, and applying optical data enhancement technology on the data and the label; on a random basis, the data is rotated between-15 and +15 degrees and scaled in the range of 0.9-1.1 therebetween; this ensures slight robustness and variability of network training;

(2) a training stage: in the training stage, parameter training is carried out on the AMU-Net network provided by the invention; learning model parameters of a training set by using a Tensorflow deep learning framework, setting the batch processing size to be 4 and the training iteration number to be 30000 by using an Adam optimizer; selecting 50% of data in a training set as a verification set, training the data by using a cross entropy loss function, and selecting a weight coefficient when the verification set is minimally lost as a final training weight in a training stage in an experiment to provide a weight parameter for a subsequent testing stage;

(3) and (3) a testing stage: randomly selecting 10% of training volume data in the MM-WHS2017 data set as a test set; firstly, a CT image with the size of 256 multiplied by 16 is directly input into a testing stage, and the final weight parameters obtained in a training stage are used for carrying out heart substructure segmentation on the testing image, so that a segmented heart substructure result graph is finally obtained.