摘要
MRI成像因其无电离辐射、软组织对比度较好等优点被广泛应用,而它最大弊端是K空间数据采样时间过长。随着压缩感知理论的形成和发展,诞生了许多结合压缩感知理论的MRI图像重构方法。这些方法可以在相同的降采样条件下,获得更高的重构质量。在基于字典学习的MRI图像重构的研究基础上,加入了用于提取图像特征的图像分割算法及全变分惩罚项。实验结果表明,可以在基于字典学习的图像重构算法基础上将重构图像的峰值信噪比提升10%~20%。
Because of its non-ionizing radiation and soft tissue contrast,MRI imaging is widely used,but the biggest drawback is that the sampling time of K-space data is too long.With the formation and development of compressed sensing theory,many MRI image reconstruction methods that combined with compressed sensing theory have been born.These methods can achieve higher reconstruction quality under the same down-sampling conditions.Based on the research of MRI image reconstruction by dictionary learning,this paper adds the image segmentation algorithm and the total variation penalty item for extracting image features.Experimental results show that the peak signal to noise ratio of reconstructed images can be improved by 10%~20% on the basis of dictionary learning algorithm.
作者
傅雪
刘文波
Fu Xue;Liu Wenbo(College of Automation Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211100,China)
出处
《电子测量技术》
2018年第11期94-98,共5页
Electronic Measurement Technology
关键词
压缩感知
字典学习
图像分割
图像重构
核磁共振图像
compressed sensing
dictionary learning
image segmentation
image reconstruction
magnetic resonance image
