电工技术学报  2024, Vol. 39 Issue (2): 343-355    DOI: 10.19595/j.cnki.1000-6753.tces.221991
电工理论 |
一种基于线性零磁场的动脉血管扫描成像方法仿真
杨丹1,2, 王雨忱1,2, 李天兆1,2, 徐彬3, 吴莹1,2
1.东北大学信息科学与工程学院 沈阳 110819;
2.智能工业数据解析与优化教育部重点实验室(东北大学) 沈阳 110819;
3.东北大学计算机科学与工程学院 沈阳 110169
Simulation of an Arterial Scanning Imaging Method Based on Linear Zero Magnetic Field
Yang Dan1,2, Wang Yuchen1,2, Li Tianzhao1,2, Xu Bin3, Wu Ying1,2
1. College of Information Science and Engineering Northeastern University Shenyang 110819 China;
2. Ministry of Education Key Laboratory of Data Analytics and Optimization for Smart Industry Northeastern University Shenyang 110819 China;
3. School of Computer Science and Engineering Northeastern University Shenyang 110169 China
全文: PDF (4411 KB)   HTML
输出: BibTeX | EndNote (RIS)      
摘要 基于磁电耦合效应的血流检测及血管成像是实现心血管疾病早期诊疗的有效方法之一。该文基于磁场与血流耦合电效应,设计一种用于动脉血管扫描成像的组合线圈结构,产生带有零磁场线(FFL)区域的线性梯度磁场;在此结构的基础上,通过控制激励电流驱动FFL实现成像区域双向扫描;结合卷积神经网络(CNN)实现磁电耦合信号与血管信息的非线性映射,进而提出一种基于线性零磁场的动脉血管扫描成像新方法。采用多物理场仿真软件COMSOL对基于线性零磁场的血管扫描成像方法进行建模,求解磁电耦合信号,验证了所提出方法的合理性和有效性。结果表明,线性梯度磁场模式下的磁电耦合信号含有血管位置、半径等信息;CNN重建血管位置误差平均值为1.569 4 mm,重建血管半径的方均误差(MSE)和相关系数(CC)平均值分别为0.054 8和0.987 0。研究结果可用于血管成像装置设计及后续相关临床应用提供研究支撑。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
杨丹
王雨忱
李天兆
徐彬
吴莹
关键词 心血管疾病磁场与血流耦合电效应零磁场线线性梯度磁场卷积神经网络COMSOL    
Abstract:Cardiovascular diseases, such as coronary artery stenosis and coronary heart disease, have become prominent in human health. Nowadays, existing diagnostic techniques in clinical practice are unsuitable for early cardiovascular disease prediction and monitoring due to trauma, high cost, radiation problems, and operation complexity. In this paper, the coupling electric effect of magnetic field and blood flow is studied, and slow imaging speed caused by an extensive range of uniform magnetic fields is considered. Then, a linear gradient magnetic field for arterial scanning imaging method is proposed.
Firstly, the topology of a combined coil for arterial vascular scanning imaging is proposed for generating a linear gradient magnetic field with a field-free line (FFL) configuration. Secondly, FFL is moved by adjusting the excitation current to realize a two-directional electronic scan of the imaging region. In addition, the curved FFL scanning trajectory is corrected. Then, the numerical simulation model of arterial scanning imaging is established by COMSOL, and the voltage signals of magnetoelectric coupling are solved by the finite element method (FEM). Finally, convolutional neural networks (CNN) are used to realize the nonlinear mapping between the magnetoelectric coupled signals and the vascular information. The coordinates of the center position and the radius of the blood vessel are obtained to reconstruct the arterial vascular image.
When the current amplitude of the gradient coil changes from 8 A to 30 A, the peak value of the magnetic field intensity in the Y-axis z direction varies from 0.024 T to 0.089 T, and the magnetic field gradient increases accordingly. The FFL is generated by superimposing the alternating driving field and the linear gradient magnetic field line. Injecting cosine alternating currents into the driving coil, FFL is electronically scanned in the yz-plane. The translated range of FFL is -30 mm to 40 mm in the y-direction and -20 mm to 25 mm in the z-direction. The CNN is trained by simulation datasets, and nonlinear mapping between the induced voltage signal and the vascular information is established. The average error of the vessel position reconstructed is 1.569 4 mm. The mean squared error (MSE) and correlation coefficient (CC) of the reconstructed vascular radius are 0.054 8 and 0.987 0, respectively.
The proposed topology of the combined coil for generating FFL has shown the potential advantage in real-time imaging of arteries. By adopting the correction coefficient in the gradient coil current, the FFL offset height can be dynamically compensated, and the linear correction of the FFL trajectory has been realized. Using CNN, a nonlinear mapping between the magneto-electric coupling signal of blood flow in linear gradient zero field excitation and the vascular information can be established, thereby obtaining the arterial vessel profile image. However, in the practical design of the imaging system, several factors must be considered, such as coil turns, coil group geometry, FFL translating the field of view (FOV), imaging speed, and image resolution. Further studies will focus on adjusting the position of each coil or optimizing the structure of combined coils for the clinical requirements of blood vessel images.
Key wordsCardiovascular disease    magnetic field and blood flow coupling electrical effect    field free line    linear gradient magnetic field    convolutional neural networks    COMSOL   
收稿日期: 2022-10-18     
PACS: TM12  
基金资助:国家自然科学基金项目(U22A20221, 61836011, 71790614)、教育部中央高校基础科研业务费项目(2020GFZD008)、辽宁省自然科学基金资助项目(2021-MS093, 2022-MS-119, 2021-BS-054)、辽宁省教育厅基础科学研究项目2021(LJKZ0014)和111项目(B16009)资助
通讯作者: 杨丹, 女,1979年生,博士,副教授,研究方向为生物电磁检测及成像。E-mail: yangdan@mail.neu.edu.cn   
作者简介: 王雨忱, 女,1998年生,硕士研究生,研究方向为电磁探测和成像技术。E-mail: 2070802@stu.neu.edu.cn
引用本文:   
杨丹, 王雨忱, 李天兆, 徐彬, 吴莹. 一种基于线性零磁场的动脉血管扫描成像方法仿真[J]. 电工技术学报, 2024, 39(2): 343-355. Yang Dan, Wang Yuchen, Li Tianzhao, Xu Bin, Wu Ying. Simulation of an Arterial Scanning Imaging Method Based on Linear Zero Magnetic Field. Transactions of China Electrotechnical Society, 2024, 39(2): 343-355.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.221991          https://dgjsxb.ces-transaction.com/CN/Y2024/V39/I2/343