| 多尺度特征融合的脊柱X线图像分割方法 |
| |
| 引用本文： | 赵阳,张俊华.多尺度特征融合的脊柱X线图像分割方法[J].计算机工程与应用,2021,57(8):214-219. |
| |
| 作者姓名： | 赵阳 张俊华 |
| |
| 作者单位： | 云南大学 信息学院，昆明 650500 |
| |
| 摘 要： | 为了精确地从X线图像中分割脊柱，提出了一种基于深度学习的脊柱X线图像分割方法，使用基于多尺度特征融合的U-Net网络进行分割。将U-Net模型中的卷积层替换成类Inception网络来提取不同尺度的特征，并进行多尺度融合。同时在跳跃连接前增加残差连接层，并在首次上采样前添加卷积块注意力模块。该模型对20幅脊柱X线图像进行验证，Dice系数为0.845 7，与近期X线脊柱图像分割方法相比，提高了0.135 1。
|
| 关 键 词： | 脊柱X线图像 U-Net 图像分割 卷积块注意力模块 |
Multi-scale Feature Fusion Method for Spinal X-Ray Image Segmentation |
| |
| ZHAO Yang,ZHANG Junhua.Multi-scale Feature Fusion Method for Spinal X-Ray Image Segmentation[J].Computer Engineering and Applications,2021,57(8):214-219. |
| |
| Authors: | ZHAO Yang ZHANG Junhua |
| |
| Affiliation: | School of Information, Yunnan University, Kunming 650500, China |
| |
| Abstract: | In order to segment spine accurately from X-ray images, a method of spine X-ray image segmentation based on deep learning is proposed, which uses U-Net network based on multi-scale feature fusion. The convolutional layer in the U-Net model is replaced with an Inception network to extract features of different scales and perform multi-scale fusion. At the same time, the residual connection layer is added in front of the skip connection, and the convolution block attention module is added in front of the first up-sampling layer. In this paper, 20 spinal X-ray images are verified by the model, and the Dice coefficient is 0.8457, which is 0.1351 higher than the recent method of spinal X-ray image segmentation. |
| |
| Keywords: | spinal X-ray image U-Net image segmentation convolution block attention module |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《计算机工程与应用》浏览原始摘要信息 |
|
点击此处可从《计算机工程与应用》下载全文 |
|
