MRI梯度控制器的设计与实现
摘要
磁共振成像(magnetic resonance imaging，MRI)是集物理学、计算机技术、电子技术、机械制造、精细化工于一体的高新技术产物，作为一种新兴的临床医疗诊断手段，它越来越显示出广阔的发展前景。谱仪是MRI的核心部件，控制着整个MRI的工作，按照一定的时序产生X轴、Y轴、Z轴三个方向的梯度场，输出相应的脉冲序列，接收和处理由于核磁共振现象而自主产生的磁共振信号，根据磁共振信号(FID或ECHO)的强度大小形成可供医疗诊断的医学图像信息。
本文在介绍MRI的原理和结构的基础上，着重研究和分析了梯度控制系统的工作原理和系统构成，给出了具体的实现方案，提供了计算各种参数的理论依据和实际的设计结果。
梯度控制系统是在计算机的控制下，为系统提供线性度满足要求的、可快速开关的梯度场，以便动态地修改主磁场，实现成像体素的空间定位。MRI成像过程中梯度控制器的工作过程是：计算机根据不同的扫描序列，将对应的梯度电流的波形按时间采样，将所得的序列以文件形式存于计算机内；初始化时，将序列值送入梯度存储器中，开始扫描时启动地址计数器，顺序选通梯度数据存储单元，确定X轴、Y轴、Z轴三个方向的梯度场时序和逻辑梯度值，同时确定不同扫描方向空间定位的旋转角度，经接口电路加载到梯度波形发生器，由梯度波形发主器中的数值计算单元中的乘法器来计算变换矩阵，然后由这些矩阵来指定梯度旋转、线性梯度定标及决定相位编码方向。
本设计中梯度控制系统通过ISA接口电路与计算机建立联
接，梯度控制器中的波形存储单元，采用高速 ZK X 16双口 RAM，
接收从主计算机订C机）传递过来的梯度波形逻辑值。这些波形
数据由序列发生单元（本设计采用Xillinx公司的大规模可编程阵
列FPGA（XC25150PQ208－5》控制取出，这些值再和变换矩阵相
乘（本设计乘法器采用＊公司的 rfMS320C5402DSP和 FPGA
（XC25150PQ208－5）共同构成人乘积值就是梯度实际控制信号，
存储在FIFO上，在序列发生单元控制下送往D／A转换器，经D／
A转换器变换成模拟信号送往梯度放大器。本系统D／AC选用
DAC16，为16位二进制输人。
整个设计电路包括序列发生、波形存贮、数值计算（进行变换
矩阵计算、矩阵乘法）、D／A转换器等几个单元。本论文绘出了用
FPGA和DSP实现的具体方案，给出了部分原理框图和一些关键
模块程序。对D／A转换器的选择标准及误差情况进行了分析。
从程序分段调试和输人、输出实验过程来看：本设计思路是正确
的，实现方案是可行的，结果表明达到了设计要求。
由于篇幅的限制，本文仅对各部分电路的硬件原理和设计进
行了适当的介绍，至于详细的DSP程序清单，VHDL程序清单，硬
件原理图以及PCB板图等未包含在本论文中。
MRI ( Magnetic Resonance Imaging) is a new and high-tech production which combines with physics computer technology elec-tronic technology machine manufacture and fine chemical. As a new medical diagnosis method, MRI has been shown out its amplitude ap-plication foreground. As a kernel of the MRI, spectrum console con-trols the operation of the MRI. It generates three gradient fields direc-ting X Y and Z respectively and emits pulse sequence to the emitting coil. It can also receive and process the magnetic resonance signal (FID, free inducing decay or ECHO, echo sequence) caused by the phenomena of magnetic resonance. According to the intension of the magnetic resonance signal, it can be used to format image applied to medical diagnosis.
In this paper, I pay my emphases on analyzing the theory and structure of the Gradient Control system under the introducing MRI general theory and structure and also present realization method in de-tail. At the same time, I provide the theory base for calculating varies of parameters and final design result.
Gradient control system controlled by the computer provides a gradient field that can be switched quickly and has well linearity. Stat-ic field can be modified automatically and spatial position of voxel ( volume element) can be provided. In imaging process, the computer samples homologous waveform of gradient current according to the scan
sequence and store pulse sequence in a file. When initial sequence start the pulse sequence values are stored in memory. When the scan is started address counters are operated and gradient data store are se-lected orderly . The computer generates timing and logic gradient of the X,Y and Z - axis according to the scan sequence and calculates rotation angles of space - oriented . Then it is transferred to gradient waveform generator to calculate the transformation matrices by multi-plier . The transformation matrices assign gradient rotation, gradient linearity and phase encode orientation .
In my paper, gradient control system communicates with the com-puter through ISA interface. The waveforms stored in gradient control system adopted higher - speed dual port RAM receive logic values of gradient waveform from the computer. We use a TMS320C5402as a DSP and a FPGA ( XC2S150PQ208 - 5 ). The multiplied values are actual gradient control signal values. These values are stored in FIFO, then are sent to D/AC under controlling by gradient control system. The values are converted to analogue signal which is sent to gradient power amplifier. In this design , the D/AC input is 16bits binary val-ue.
In the circuit, it consists of waveform store , matrix multipliers , matrix calculator and sequencers . This paper also provides many structure diagrams, simulation wave figures and some VHDL codes. It also gives selective standard and offset analysis for D/AC . The experi-ment process and stimulator results indicate that this design is correct and plan is available . The testing result indicates that this design can meet the need of the project.
Because the room is limited, I pay more emphases on researching and analyzing the principle and structure of all parts. The detail DSP
sequence list, VHDL sequence list , hardware principle scheme and PCB scheme are not in this paper.
本文在介绍MRI的原理和结构的基础上，着重研究和分析了梯度控制系统的工作原理和系统构成，给出了具体的实现方案，提供了计算各种参数的理论依据和实际的设计结果。
梯度控制系统是在计算机的控制下，为系统提供线性度满足要求的、可快速开关的梯度场，以便动态地修改主磁场，实现成像体素的空间定位。MRI成像过程中梯度控制器的工作过程是：计算机根据不同的扫描序列，将对应的梯度电流的波形按时间采样，将所得的序列以文件形式存于计算机内；初始化时，将序列值送入梯度存储器中，开始扫描时启动地址计数器，顺序选通梯度数据存储单元，确定X轴、Y轴、Z轴三个方向的梯度场时序和逻辑梯度值，同时确定不同扫描方向空间定位的旋转角度，经接口电路加载到梯度波形发生器，由梯度波形发主器中的数值计算单元中的乘法器来计算变换矩阵，然后由这些矩阵来指定梯度旋转、线性梯度定标及决定相位编码方向。
本设计中梯度控制系统通过ISA接口电路与计算机建立联
接，梯度控制器中的波形存储单元，采用高速 ZK X 16双口 RAM，
接收从主计算机订C机）传递过来的梯度波形逻辑值。这些波形
数据由序列发生单元（本设计采用Xillinx公司的大规模可编程阵
列FPGA（XC25150PQ208－5》控制取出，这些值再和变换矩阵相
乘（本设计乘法器采用＊公司的 rfMS320C5402DSP和 FPGA
（XC25150PQ208－5）共同构成人乘积值就是梯度实际控制信号，
存储在FIFO上，在序列发生单元控制下送往D／A转换器，经D／
A转换器变换成模拟信号送往梯度放大器。本系统D／AC选用
DAC16，为16位二进制输人。
整个设计电路包括序列发生、波形存贮、数值计算（进行变换
矩阵计算、矩阵乘法）、D／A转换器等几个单元。本论文绘出了用
FPGA和DSP实现的具体方案，给出了部分原理框图和一些关键
模块程序。对D／A转换器的选择标准及误差情况进行了分析。
从程序分段调试和输人、输出实验过程来看：本设计思路是正确
的，实现方案是可行的，结果表明达到了设计要求。
由于篇幅的限制，本文仅对各部分电路的硬件原理和设计进
行了适当的介绍，至于详细的DSP程序清单，VHDL程序清单，硬
件原理图以及PCB板图等未包含在本论文中。
MRI ( Magnetic Resonance Imaging) is a new and high-tech production which combines with physics computer technology elec-tronic technology machine manufacture and fine chemical. As a new medical diagnosis method, MRI has been shown out its amplitude ap-plication foreground. As a kernel of the MRI, spectrum console con-trols the operation of the MRI. It generates three gradient fields direc-ting X Y and Z respectively and emits pulse sequence to the emitting coil. It can also receive and process the magnetic resonance signal (FID, free inducing decay or ECHO, echo sequence) caused by the phenomena of magnetic resonance. According to the intension of the magnetic resonance signal, it can be used to format image applied to medical diagnosis.
In this paper, I pay my emphases on analyzing the theory and structure of the Gradient Control system under the introducing MRI general theory and structure and also present realization method in de-tail. At the same time, I provide the theory base for calculating varies of parameters and final design result.
Gradient control system controlled by the computer provides a gradient field that can be switched quickly and has well linearity. Stat-ic field can be modified automatically and spatial position of voxel ( volume element) can be provided. In imaging process, the computer samples homologous waveform of gradient current according to the scan
sequence and store pulse sequence in a file. When initial sequence start the pulse sequence values are stored in memory. When the scan is started address counters are operated and gradient data store are se-lected orderly . The computer generates timing and logic gradient of the X,Y and Z - axis according to the scan sequence and calculates rotation angles of space - oriented . Then it is transferred to gradient waveform generator to calculate the transformation matrices by multi-plier . The transformation matrices assign gradient rotation, gradient linearity and phase encode orientation .
In my paper, gradient control system communicates with the com-puter through ISA interface. The waveforms stored in gradient control system adopted higher - speed dual port RAM receive logic values of gradient waveform from the computer. We use a TMS320C5402as a DSP and a FPGA ( XC2S150PQ208 - 5 ). The multiplied values are actual gradient control signal values. These values are stored in FIFO, then are sent to D/AC under controlling by gradient control system. The values are converted to analogue signal which is sent to gradient power amplifier. In this design , the D/AC input is 16bits binary val-ue.
In the circuit, it consists of waveform store , matrix multipliers , matrix calculator and sequencers . This paper also provides many structure diagrams, simulation wave figures and some VHDL codes. It also gives selective standard and offset analysis for D/AC . The experi-ment process and stimulator results indicate that this design is correct and plan is available . The testing result indicates that this design can meet the need of the project.
Because the room is limited, I pay more emphases on researching and analyzing the principle and structure of all parts. The detail DSP
sequence list, VHDL sequence list , hardware principle scheme and PCB scheme are not in this paper.
引文
[1] 赵喜平．磁共振成像系统的原理及其应用．北京．科学出版社，2000．12
[2] 高上凯．医学成像系统．北京．清华大学出版社，2000．3
[3] 徐志军 徐光辉．CPLD／FPGA的开发与应用．北京．电子工业出版社，2002．1
[4] 程佩青．数字信号处理教程(第二版)．北京．清华大学出版社，2001．8
[5] 吴世法．近代成像技术与图像处理．北京．国防工业出版社，1997．3
[6] 李景华 杜玉远．可编程逻辑器件与EDA技术．沈阳．东北大学出版社，2002．3
[7] 孙廷才 王杰 孙中健．工业控制计算机组成原理．北京．清华大学出版社，2001．5
[8] Ray H. Hashemi. MRI The Basics. Williams & Wilkins company, 1997 Using Root Moments. IEEE TRANSACTIONS ON Circuits and Systems, June 2001 volume 48 number6
[9] I. Ando and G. A. Webb. Theory of NMR Parameters. London Academic Press, 1983
[10] G. Bradley, Jr. J. S. Tsuruda. MR sequence parameter optimization: an algorithmic approach. AJR 1987,149:815-823
[11] J.A. Patton. MR imaging instrumentation aand image artifacts. RadioGraphics, 1994,14:1083-1096
[12] P. M. Parizel, A. M. De Schepper. low-field MRI makes economic progress. Diagnostic Imaging Europe 1996:Oct
[13] TMS320C54x Peripherals Reference Guide. Texas Instruments
Incorporated, 1999
[14] TMS320C54x CPU and Instruction Set Reference Guide. Texas Instruments Incorporated, 2000
[15] TMS320C54x Programmer's Guide. Texas Instruments Incorporated,2000
[16] TMS320C54x C Source Debugger. Texas Instruments Incorporated, 1998
[17] TMS320C54x Assembly Language Tools. Texas Instruments Incorporated,2000
[18] TMS320C54x Tools: Vector Table and Boot ROM Creation. Texas Instruments Incorporated, 1999
[19] TMS320C54x McBSP Initialization. Texas Instruments Incorporated, 1998
[20] TMS320C54x Board Design Consideration for Debug. Texas Instruments Incorporated, 1999
[21] TMS320C54x DMA Example Application. Texas Instruments Incorporated, 1999
[22] TMS320C54x Peripheral Support Library programmer's Reference. Texas Instruments Incorporated, 1998
[2] 高上凯．医学成像系统．北京．清华大学出版社，2000．3
[3] 徐志军 徐光辉．CPLD／FPGA的开发与应用．北京．电子工业出版社，2002．1
[4] 程佩青．数字信号处理教程(第二版)．北京．清华大学出版社，2001．8
[5] 吴世法．近代成像技术与图像处理．北京．国防工业出版社，1997．3
[6] 李景华 杜玉远．可编程逻辑器件与EDA技术．沈阳．东北大学出版社，2002．3
[7] 孙廷才 王杰 孙中健．工业控制计算机组成原理．北京．清华大学出版社，2001．5
[8] Ray H. Hashemi. MRI The Basics. Williams & Wilkins company, 1997 Using Root Moments. IEEE TRANSACTIONS ON Circuits and Systems, June 2001 volume 48 number6
[9] I. Ando and G. A. Webb. Theory of NMR Parameters. London Academic Press, 1983
[10] G. Bradley, Jr. J. S. Tsuruda. MR sequence parameter optimization: an algorithmic approach. AJR 1987,149:815-823
[11] J.A. Patton. MR imaging instrumentation aand image artifacts. RadioGraphics, 1994,14:1083-1096
[12] P. M. Parizel, A. M. De Schepper. low-field MRI makes economic progress. Diagnostic Imaging Europe 1996:Oct
[13] TMS320C54x Peripherals Reference Guide. Texas Instruments
Incorporated, 1999
[14] TMS320C54x CPU and Instruction Set Reference Guide. Texas Instruments Incorporated, 2000
[15] TMS320C54x Programmer's Guide. Texas Instruments Incorporated,2000
[16] TMS320C54x C Source Debugger. Texas Instruments Incorporated, 1998
[17] TMS320C54x Assembly Language Tools. Texas Instruments Incorporated,2000
[18] TMS320C54x Tools: Vector Table and Boot ROM Creation. Texas Instruments Incorporated, 1999
[19] TMS320C54x McBSP Initialization. Texas Instruments Incorporated, 1998
[20] TMS320C54x Board Design Consideration for Debug. Texas Instruments Incorporated, 1999
[21] TMS320C54x DMA Example Application. Texas Instruments Incorporated, 1999
[22] TMS320C54x Peripheral Support Library programmer's Reference. Texas Instruments Incorporated, 1998