论文部分内容阅读
为精确地进行并行磁共振成像,文章利用字典学习的强大捕捉细节和稀疏开发能力,提出了一种基于自适应稀疏表达的重建方法。该方法将并行磁共振重建问题转化为最小化由字典学习和数据拟合项构成的目标函数,并采用了分而治之的方案求解未知变量。为验证其有效性,将该方法与目前主流的两种方法在人体实际磁共振数据上进行了测试。测试结果显示,文章提出的方法能在抑制图像噪声的同时较好地保存图像细节。
In order to carry out parallel magnetic resonance imaging accurately, this paper proposes a reconstruction method based on adaptive sparse representation, which utilizes the powerful capturing details and sparse development ability of dictionary learning. The method transforms the problem of parallel magnetic resonance reconstruction into minimizing the objective function composed of dictionary learning and data fitting, and uses the divide and conquer scheme to solve the unknown variable. In order to verify its validity, this method and the current two methods are tested on the actual human magnetic resonance data. The test results show that the proposed method can save the image detail well while suppressing image noise.