波谱学杂志 ›› 2022, Vol. 39 ›› Issue (1): 33-42.doi: 10.11938/cjmr20212903

• 研究论文 • 上一篇    下一篇

基于神经网络拟合的腹部化学交换饱和转移成像

王志超1,张记磊2,赵羽3,华婷4,汤光宇4,李建奇1,*()   

  1. 1. 华东师范大学 物理与电子科学学院, 上海市磁共振重点实验室, 上海 200062
    2. 飞利浦医疗公司, 上海 200040
    3. Institute of Imaging Science, Vanderbilt University, Tennessee 37232
    4. 上海市第十人民医院放射科, 上海 200072
  • 收稿日期:2021-03-30 出版日期:2022-03-05 发布日期:2021-05-16
  • 通讯作者: 李建奇 E-mail:jqli@phy.ecnu.edu.cn
  • 基金资助:
    华东师范大学“幸福之花”基金资助项目

CEST Imaging of the Abdomen with Neural Network Fitting

Zhi-chao WANG1,Ji-lei ZHANG2,Yu ZHAO3,Ting HUA4,Guang-yu TANG4,Jian-qi LI1,*()   

  1. 1. Shanghai Key Laboratory of Magnetic Resonance, School of Physics and Electronic Science, East China Normal University, Shanghai 200062, China
    2. Philips Healthcare, Shanghai 200040, China
    3. Institute of Imaging Science, Vanderbilt University, Tennessee 37232, USA
    4. Department of Radiology, Shanghai Tenth People's Hospital, Shanghai 200072, China
  • Received:2021-03-30 Online:2022-03-05 Published:2021-05-16
  • Contact: Jian-qi LI E-mail:jqli@phy.ecnu.edu.cn

摘要:

磁共振成像(Magnetic Resonance Imaging,MRI)化学交换饱和转移(Chemical Exchange Saturation Transfer,CEST)技术在临床诊断中展现了巨大的潜力,但在腹部成像中受到主磁场偏移量大的挑战,而且利用传统的非对称性分析法得到的酰胺质子转移(Amide Proton Transfer,APT)成像对比度受到核奥氏增强(Nuclear Overhauser Enhancement,NOE)效应的干扰.本文提出了一种基于神经网络拟合的CEST后处理方法,对每个像素采集得到的Z谱特征进行识别,不需要额外序列扫描即可得到背景参考Z谱与主磁场偏移量,用以校正和获得理想的Z谱,并进一步分离得到源自APT效应与NOE效应的信号.鸡蛋清和健康志愿者腹部成像结果显示,本文提出的基于神经网络的CEST后处理方法效果较好.

关键词: 磁共振成像(MRI), 化学交换饱和转移(CEST), 酰胺质子转移(APT), 核奥氏增强(NOE), 神经网络

Abstract:

Chemical exchange saturation transfer (CEST) imaging shows great potential in clinical applications. However, CEST imaging is challenging in abdomen due to the large B0 shift. Meanwhile, the nuclear Overhauser enhancement (NOE) effect contaminates amide proton transfer (APT) image when using the conventional asymmetry analysis. In this paper, a CEST post-processing approach based on neural network fitting was proposed. Through recognizing the characteristics of acquired Z-spectrum, the background reference Z-spectrum and the B0offset were obtained and used to correct the acquired Z-spectrum. The APT effect and NOE effect could be calculated by subtracting the background reference Z-spectrum from the acquired Z-spectrum. The proposed CEST post-processing approach was validated by the egg white imaging and the abdomen imaging on four healthy volunteers.

Key words: magnetic resonance imaging (MRI), chemical exchange saturation transfer (CEST), amide proton transfer (APT), nuclear Overhauser enhancement (NOE), neural network

中图分类号: