Next-Generation MRI Scanner

Transcription

1 Next-Generation MRI Scanner Final Project Report, Spring 2016 Full Report Prepared by: Erika Bratschun : E.E. Undergraduate Daniel Tait : E.E. Undergraduate Nabeel Moin : E.E. Undergraduate Department of Electrical and Computer Engineering Colorado State University Fort Collins, CO Project advisor(s): Branislav Notaros, Professor Pranav Athalye, Ph.D. Student Approved By : Branislav Notaros

2 Abstract There are over twenty-five thousand Magnetic Resonance Imaging (MRI) machines being used around the world for medical imaging. Existing problems with modern MRI machines include insufficient resolution which makes seeing the subtle difference in the soft tissue being analyzed difficult, and the lengthy duration of scans, sometimes taking as long as two hours. The goals of this project are to improve the time of the scans and the clarity of the images produced by MRI scanners. Specifically, the project has researched improvements to the radiofrequency (RF) coils, which are an integral component of MRI scanners. The proposed design is intended for use in 7T scanners and consists of a quadrifilar helix with a grounding backplate excited in quadrature at 300 MHz, which differs greatly from the current industry standard of a birdcage-shaped coil. It has been found from preliminary research that undesired eddy currents are produced within the grounding backplate of the RF coil, which distort the produced images. As a result, the foremost goal of this project thus far has been the elimination of the eddy currents from previous RF coil designs. The other goal of the project, speeding up scan times, has been achieved by employing multiple independent coils instead of one singularly excited coil, which opens up the possibility for parallel imaging. The first step in the approach towards achieving these goals has been through the application of High Frequency Structural Simulator (HFSS), a commercial software used for modeling electromagnetic structures, in which the RF coils were designed and performance was optimized. Then, the designs were fabricated and tested at the RF lab at Broadcom. In the future, it is planned for the designs to undergo more extensive testing at the Center for Magnetic Resonance Research in Minnesota. 1

3 Table of Contents Abstract... 1 Table of Figures Introduction Magnetic Resonance Imaging Project Description Project Specifications and Design Constraints Report Overview Research and Theory Larmor Precession Nuclear Magnetic Resonance Components of MRI Scanners Pulse Sequences and Signal Processing Imaging Phantoms Factors in MRI Quality Eddy Currents Metamaterials in MRI Simulating Eddy Currents Uniformity Modeling in MATLAB Summary of Previous Work Previous Design Eddy Current Artifacts Hardware Design Preliminary Backplate Designs Final Backplate Design Variations in Phantom Distance Variations in Phantom Shape Variations in Helix Diameter Designs Involving Straight Wire Antennas Designs Involving Multiple Helices Coil Fabrication Hardware Testing

4 5. Conclusions and Future Work Conclusion Further Testing Recommendation for Continuation References Appendix A: Abbreviations Appendix B: Budget Appendix C: Revision of Timelines and Deliverables Appendix D: Code Listing

5 Table of Figures Figure 1 Diagram illustrating eddy currents Figure 2 Eddy current simulation model in ANSYS Maxwell Figure 3 Eddy currents in mesh backplate Figure 4 Eddy currents in solid backplate Figure 5 Eddy current vectors in solid backplate Figure 6 Coil model for MATLAB simulation Figure 7 2D field plots from HFSS (left) and MATLAB (right) Figure 8 Previously designed RF coil Figure 9 HFSS model for old design Figure 10 CMRR experiment (top) vs. HFSS simulation (bottom) Figure 11 Current design: 60 cm quadrifilar helical coil Figure 12 Axial magnetic field plot for initial design Figure 13 Backplate with radial slots Figure 14 Backplate with azimuthal slots Figure 15 Meshed backplate, 24 radial slots Figure 16 Meshed backplate, 28 cm diameter hole Figure 17 Meshed backplate, 30 cm diameter hole Figure 18 Minimal surface area design Figure 19 Final surface area-minimized design Figure 20 Axial magnetic field plot for final surface area-minimized design Figure 21 1D plot comparing eddy currents along a radial cut of original and mesh backplate.. 22 Figure 22 Backplate notches cut along outermost ring (left) and then loops removed (right) Figure 23 Backplate with removal of loops and reduced size Figure 24 Backplate with removal of loops and less connections around the perimeter Figure 25 "Sun" backplate Figure 26 1D axial magnetic field pattern solid backplate (dashed) and sun backplate (solid) Figure 27 2D Field, HRCP plot of the final backplate design Figure 28 1D axial magnetic field plot of solid backplate (dashed) vs. sun backplate with thicker strips (solid) Figure 29 Phantom 815 mm from bakplate (default) Figure 30 Axial magnetic field for phantom 815mm from backplate

6 Figure 31 Phantom 586 mm from backplate Figure 32 Axial magnetic field for phantom 586 mm from backplate Figure 33 Phantom 725mm from backplate Figure 34 Axial magnetic field for phantom 725mm from backplate Figure 35 Phantom with convex ridges Figure 36 1D axial magnetic field for convex phantom Figure 37 Phantom with concave notches Figure 38 1D axial magnetic field for concave phantom Figure 39 Phantom with sinusoidal shape Figure 40 1D axial magnetic field for sinusoidal phantom Figure 41 60cm long, 36cm diameter coil Figure 42 1D axial magnetic field for alternate coil size Figure cm diamter coil Figure 44 1D axial magnetic field for 40 cm diameter coil Figure 45 Straight-wire, single backplate model (left) and axial magnetic field (right) Figure 46 Straight-wire, double backplate model (left) and axial magnetic field (right) Figure 47 2D field plot of HRCP (left) and HLCP (right) for straight wire design Figure 48 Multiple helix cascade (left) and 2D HRCP plot (right) Figure 49 Other multi-helix designs Figure 50 Fabricated coil with labelled materials Figure 51 HFSS simulation (top) and Broadcom experiment (bottom) for 1-channel excitation 38 Figure 52 Smith chart for a single channel

7 1. Introduction 1.1. Magnetic Resonance Imaging Magnetic resonance imaging (MRI) is an imaging technique which exploits the principle of Larmor precession and nuclear magnetic resonance (NMR). Larmor precession refers to the precession of the magnetic moment of a proton when exposed to a magnetic field, while nuclear magnetic resonance is a phenomenon where precessing protons will produce detectable resonations when introduced to radio waves of the same frequency. This imaging technique originates from experiments done in 1973 by Paul Lauterbur, who first produced the images through the use of NMR. Years later, the idea of using linear magnetic gradients to gather information from an object led to the development of MRI scanners widely used today. MRI also has a number of advantages over other imaging techniques. MRI scanners are safer than methods which employ harmful X-rays, which may potentially increase the risk of cancer and cause harm to the human body, and excels at producing images of soft tissue. The widespread use of these medical devices has extended the lives of countless people, thus warranting any research that could potentially contribute towards the further improvement of this technology. Modern MRI scanners consist of magnetic, gradient, and radiofrequency (RF) coils. This project seeks to make contributions towards this technology by improving the RF coils. The RF coils, which are responsible for both exciting protons and receiving the signal are crucial to the quality of the images produced through the MRI process. Improved designs of the RF coils used in MRI scanners will provide better image quality, providing doctors with clearer and more accurate information about their patients Project Description This research and design project aims to make improvements to the RF coils used in 7.0T MRI scanners. The first stages of the project have involved simulating designs in software, namely HFSS, WIPL-D, and Maxwell, to test field patterns, S-parameters, and eddy currents which has allowed for the team to optimize the design at a low cost. It is known from previous tests that the major concern with this next round of designs is the effects of eddy currents on the imaging. The eddy currents are an undesirable effect resulting from the time-varying magnetic gradients, resulting in distortions in the generated MR images. Although these eddy currents cannot be tested until the final stages of testing at the Center for Magnetic Resonance Research (CMRR), research and prior knowledge has led to a design in the backplate that will lead to a reduction of eddy currents. The other primary focus is sizing. The RF coil can be used for anything from a head scan to a full body. This is another element that must be taken into consideration for the RF coil design, because results indicate that any changes to the coil will have a large impact on the intensity of the magnetic field. The latter stages of the project included actual fabrication of RF coil designs using materials suitable for an MRI machine. The coil was then tested at the RF testing lab at Broadcom. 6

8 1.3. Project Specifications and Design Constraints Coil Length: 60cm (originally cm) Coil Diameter: 38cm (originally 55-65cm), body coil Minimum HRCP Magnitude: 100mA/m HRCP Field Tolerance: ±15% Minimum HRCP to HLCP Ratio: 10 Axial Ratio: > 1, pure Circular Polarization Low electric field Images produced by the final design should have no distortion due to eddy currents Report Overview The contents of this report are a compendium of the most important work done on this project throughout the 2016 fall and spring semesters. Within chapter 2, the findings of our research into MRI and other topics related to this project are discussed. Chapter 3 will discuss previous research done on this project and the problems which have been encountered during this research. Next, chapter 4 will discuss proposed changes to the RF coil design, as well as various findings which could potentially improve future designs. Lastly, chapter 5 will conclude this document and discuss future plans for this project. 7

9 2. Research and Theory 2.1. Larmor Precession Larmor precession occurs as a result of the application of a magnetic field to a proton. The nucleus of a hydrogen atom, which is a single proton, has a magnetic moment. When a magnetic field is introduced to a proton, its magnetic moment will begin to precess at an angular frequency directly proportional to the strength of the magnetic field introduced. Conceptually, this may be understood by relating it to how a top precesses while spinning. The directly proportional relationship between the magnetic field and the angular velocity of the magnetic moment is exploited in MRI Nuclear Magnetic Resonance MRI relies on the phenomenon of NMR, which occurs as a result of the application of RF pulses to hydrogen atoms, typically residing within soft tissue of a scanned region. When the nuclei of these hydrogen atoms are excited by these RF pulses, they will enter an excited state. If the magnetic field being applied to these atoms is considered to be directed in the z-direction, the application of the already precessing hydrogen atoms will further excite them and push them into a state where the rotation of their magnetic moment is perpendicular to the z-directed magnetic field. It should be noted that the frequency of these RF pulses is at the same frequency as the Larmor precession of the hydrogen nucleus. When the RF pulse ceases, the excited hydrogen atoms will relax. This relaxation, may be classified into both spin-lattice (T1) and spin-spin (T2) relaxation. T1 relaxation is the process of how a proton will realign with a magnetic field. The time of this recovery is predictable, as it is dependent upon the strength of the external magnetic field. T2 relaxation is the process of how neighboring protons will dephase with one another, because energy will be transferred between their spins. Both of these relaxations occur independently from one another and may be detected to produce images Components of MRI Scanners Although MRI scanners consist of a large number of components, they possess three coils which are essential for their function. These are the primary magnetic coil, gradient coils, and RF coils. The magnetic coil is a large magnetic coil which produces a very strong magnetic field, typically of 3.0T or higher. For proper functioning, this coil must be cooled with liquid helium, so that it may be superconductive. The gradient coils are three magnets responsible for creating a relatively small magnetic gradient in the x-, y-, or z-direction. The purpose of this gradient is to encode a phase and frequency into each proton within the scanned region, which are then processed into a pixelated image. Finally, the RF coils are responsible for transmitting RF pulses and receiving RF resonations. Our project has thus far investigated RF coils for use in MRI scanners which employ 7.0T magnetic fields. 8

10 2.4. Pulse Sequences and Signal Processing The RF and gradient coils both operate through the application of various pulse sequences. Although this project has, at this point, not investigated ways of improving pulse sequences; it is necessary to understand their role in the scanning process to have an accurate understanding of MRI. The pulses which are introduced to the gradient coils affect the Larmor precession of protons within this scanned region. Through a combination of pulses applied to each of the gradient coils, the phase and frequency of these protons may be encoded. Once a region, or slice, within the scanned region is selected, information may be extracted from it by applying a pulse sequence to the RF coils which excite protons within the scanned region, the relaxation of which may then be detected. The time-domain data is then processed into spectral data through the application of a Fourier transform. These points of data are what determine the grayscale value of the pixels which MR images are composed of Imaging Phantoms When initially performing tests on new MRI technology, it is not ethically sound to use humans or other living specimens as the test subjects. Therefore, in order to simulate the manner in which RF waves penetrate into conductive media (namely, the human body), a bottle filled with saline is used as a test subject instead. This specimen substitute, known as the phantom, is more convenient and ethical to use for testing the performance of MRI systems under development. These phantoms are also oftentimes included in computer simulations of MRI systems Factors in MRI Quality There are numerous factors which contribute towards the quality of the images produced by MR scans. Patients preparing for MR scans may even be administered contrasting agents before a scanning procedure, to manipulate the results of these scans. The factors of immediate interest to this project however are the field pattern and magnitude of the magnetization response of the RF coil, as well as the prevalence of eddy currents. First, the magnetization response of the RF coil plays a significant role in the image quality. The homogeneity of the magnetization response will affect the homogeneity of the received or transmitted signal. For example, a homogenous response of a homogenous material may produce a purely white grayscale MR image, while an inhomogeneous response of the same material may appear gray in some areas. Furthermore, the magnitude will determine the strength of the signal that is received from or emitted towards a certain region. Secondly, eddy currents produce undesirable artifacts within the resulting images. During an MR scan, the pulse sequences involved with the gradient coils may cause eddy currents to form on wires and other conductive media within and around the scanned region. These currents then create undesirable magnetic fields which interfere with the scanning process. These artifacts may be minimized by reducing the conductive media involved in the MR scanning process. 9

11 2.7. Eddy Currents The gradient coils produce a rapidly switching magnetic field. Because of Faraday s Law, which states that a changing magnetic field will induce a current in a conductor, the gradient coils induce currents in the RF coil ground plate. Generally, these currents resemble eddies that form in a body of water when, for example, an oar is rowed through it. These currents are not desired in MRI due to their numerous unintentional consequences. Firstly, the fields produced by the eddy currents interfere with the fields used for imaging, resulting in undesired artifacts in the image. Additionally, because the plate is a current-carrying body in a magnetic field, it will experience a Lorentz force and vibrate, causing a loud acoustic noise. This exacerbates the existing problem of loud background noises in MRI scanners, and should be avoided in order to minimize patient discomfort. Figure 1 Diagram illustrating eddy currents 2.8. Metamaterials in MRI Metamaterials are synthetic materials which have a negative relative permittivity or negative relative permeability (εr or μr). Outside research has found that the application of metamaterial linings to the bore of MR scanners may improve image quality. This suggests that metamaterials may have further applications in MRI scanners. 10

12 2.9. Simulating Eddy Currents MRI scanners are complex devices, so it is difficult to produce models which may accurately represent how their scans may induce eddy currents. A more simplified model was created to demonstrate how susceptible a material or backplate geometry may be to the induction of eddy currents. This was done through the use of ANSYS Maxwell, a software similar to ANSYS HFSS, which is capable of modeling low frequency fields. This was also modeled with the use of a 3D model, rather than a 2D model, as it was believed that this would better represent the magnetic fields within an MRI scanner. The eddy currents induced within each backplate could then be compared to one another to give an idea of their ability to suppress eddy currents. In this software, a model of a current loop was created. This model may be seen below in Figure 2. The backplate of interest was then imported into the design, translated into a 3D object, assigned a material (typically copper), and placed at the center of the current loop. The solution type used within each simulation was eddy currents. These simulations were ran at frequencies up to 10kHz, the highest operating frequency which our group believed may be responsible for the induction of eddy currents within an MRI scanner. Figure 2 Eddy current simulation model in ANSYS Maxwell The results of these simulations were examined in several ways. Typically, results were examined through 1D and 2D plots along a central slice of the backplate. Examples of these plots, for the mesh and solid back plates are shown in Figure 3 and Figure 4, respectively. As shown, the eddy currents within the mesh backplate are different those within the solid backplate. More quantitative results could be obtained through the fields calculator, found within the ANSYS software, by means of the surface and volume integral functions. Our group relied upon both qualitative and quantitative results to determine whether a backplate could suppress eddy currents. It should be noted that the higher currents around the perimeter of the backplate are partially due to their proximity to the current loop. Originally this was recognized as a flaw within our methodology, but it is believed that the z-coils within actual MRI scanners would produce in a similar phenomenon. Simulations were also performed, where the backplate was placed on its side by rotating it along the x-axis, which found that eddy currents induced by the x- and y- gradient coils were negligible. 11

13 Figure 3 Eddy currents in mesh backplate Figure 4 Eddy currents in solid backplate Figure 5 shows the results of the solid backplate eddy currents simulated at 5kHz. The arrows indicate the direction of current. This confirmed our belief in how eddy currents were induced within the backplate. Figure 5 Eddy current vectors in solid backplate 12

14 2.10. Uniformity Modeling in MATLAB One of the greatest challenges faced in this project was the task of changing the field pattern within the phantom. It was believed that by arraying several helix segments and adjusting the magnitude and phase of their excitations, that an altered and more even field pattern could be achieved. This could be done directly, by adjusting the power and phase of the excitations in ANSYS simulation results, or by exporting the field results within the phantom and importing them into MATLAB. The advantages of manipulating the data within MATLAB are that it requires significantly less time to generate a new set of data for different combinations of feed currents and the scripting language used in MATLAB is more user-friendly than the fields calculator within the ANSYS software. This method was implemented for data collected from a design similar to Figure 6, which used eight segments, rather than four. Furthermore, field patterns for results computed in ANSYS and in MATLAB may be seen below. This pattern was obtained by varying the feed currents of each helix segment. Attempts were also made to adjust the phase of each segment within the MATLAB script, but it was found that the data obtained this way did not match the results of the ANSYS software. The reason for this failure may have been due to the small inter-element spacing between each helix segment or because of error during implementation. There were also attempts to produce MATLAB code capable of seeking through data generated for many different feed current combinations (e.g. binomial or triangular). In this code, the user could define a region of interest within the phantom and the program would find the feed currents which produced the most even or high magnitude field for that region. This method was ultimately abandoned, however, due to time constraints. Further discussion on our findings on segmented helices may be seen in section 4.7. A listing of one version of the code which was written may be seen in Appendix D. 13

15 Figure 6 Coil model for MATLAB simulation Figure 7 2D field plots from HFSS (left) and MATLAB (right) 14

16 3. Summary of Previous Work 3.1. Previous Design This senior design project, although new, is a continuation of previously done research on the application of new RF coils designs in ultra-high field MRI scanners. More specifically, this research has explored applications involving helical antennas. Listed below are the specifications of the previous design and the design itself shown may be seen in Figure 8. Backplate Diameter: Coil Length: Coil Pitch: Saline Phantom Diameter: Saline Phantom Length: Saline Phantom Distance: 35cm 2m 107mm 16cm 370mm 815mm Figure 8 Previously designed RF coil Previous work has also resulted in the creation of helical antenna models for use in HFSS and WIPL-D. The designs. The HFSS model used for the old design may be seen in Figure 9. 15

17 Figure 9 HFSS model for old design 3.2. Eddy Current Artifacts The greatest flaws in this early design were the eddy current artifacts visible in the resulting MR images. This is greatly undesirable, thus the elimination of these eddy currents became one of the main objectives of our project for this semester. Test results from the Center for Magnetic Resonance Research (CMRR) at the University of Minnesota suggested that these eddy currents were the result of a solid backplate design. Eddy current from previously tested coils found to disturbance the imaging as seen in Figure 10. As mentioned, the disturbances are a results of the gradient coils producing a time varying electromagnetic field which induces the currents which are causing the visible distortions in the figures below. 16

18 Figure 10 CMRR experiment (top) vs. HFSS simulation (bottom) 4. Hardware Design The basic design of the quadrifilar helical coil has three major components: the coils, backplate, and exciters. There are four independent helical coils (hence, quadrifilar), each with their own excitation source, excited in quadrature (90 degree phase shift between adjacent coils). Each helical coil acts as a monopole antenna. The backplate simulates an antenna ground plane, making the monopole into an effective dipole, thereby increasing its radiation strength. The effectiveness of the design with respect to the field homogeneity and HRCP to HLCP was evaluated by looking at the axial magnetic field in the phantom, that is, the magnetic field along the center axis of the phantom. The axial field plot for the basic design can be seen in Figure 12 with the magnitude of HRCP in red and the magnitude of HLCP in blue. In this plot, it can be seen that the basic design has fairly good field homogeneity and a very high ratio of HRCP to HLCP. 17

19 Figure 11 Current design: 60 cm quadrifilar helical coil Figure 12 Axial magnetic field plot for initial design 4.1 Preliminary Backplate Designs Since the distortions in previously acquired images were determined to be caused by eddy currents in the backplate, it was necessary to design a new backplate that would suppress these currents and eliminate the previously discussed distortions. The steps taken towards the removal of these distortions were based on our theoretical understanding of eddy currents. Although total elimination of eddy currents would be ideal, realistically, if the magnitude of the currents was reduced significantly, the residual artifacts could be removed using image processing techniques. The positioning of the backplate in earlier tests was such that the plate was normal to the xy-plane, in the direction of the B0 field (z-directed). The position of the backplate would suggest that the changing magnetic gradient from the z-coil would be responsible for inducing eddy currents within the backplate as a result of Faraday s law. This would mean that the eddy currents from this gradient flow azimuthally along the backplate surface. However, because the backplate has thickness, there exists the possibility that the magnetic gradient from the x- and y-coils, normal upon the sides of the backplate, may also be responsible for the induction of eddy currents. If this were the case, then the possibility of currents which flow around the backplate in the x- and y- directions, around the surface of the backplate, must also be taken into consideration. Although 18

20 the difference in surface area between the side and base of the backplate would suggest that the azimuthally directed currents contribute more to the distortions, the performance requirements of our design warranted that steps should be taken towards the removal of both types of eddy currents. For this reason, investigations were made into the addition of radial and azimuthal slots, which would impede the flow of eddy currents. The initial changes made to the backplate may be seen in Figure 13 and Figure 14. Figure 13 Backplate with radial slots Figure 14 Backplate with azimuthal slots These changes to the backplate suggested that the idea of having a mesh backplate, where the backplate would be composed of wires, would be worth exploring. This resulted in the creation of a backplate design, where the backplate was made from concentric rings and wires. The initial designs for this design strategy may be seen in Figure

21 Figure 15 Meshed backplate, 24 radial slots At this point, the possibility of increasing the diameter of the central hole within the backplate was explored. Originally, it was thought that a hole in the center of the backplate negatively impacted the field within the pattern, as the grounding ability of the backplate would be reduced. However, by running tests on a backplate with a hole, it was established that the radiation pattern was not significantly impacted. For this reason, the effects of increasing the size of the hole within the backplate were investigated. Some of these designs may be seen in Figure 16 and Figure 17. Figure 16 Meshed backplate, 28 cm diameter hole Figure 17 Meshed backplate, 30 cm diameter hole It was found from these changes to the hole in the backplate that the diameter of the hole in the backplate could contribute towards field homogeneity within the phantom. In the 1D field pattern for many simulations, there are consistently results where an initial jump, or some erratic behavior within the magnitude of the field may be observed. By changing the diameter of this 20

22 central hole, this irregularity may be corrected. For this reason, any future designs where there are irregularities within the field pattern of the phantom may be corrected by making changes to the central hole within the backplate if these designs are used. The final size of this hole was chosen based on the horizontal, sinusoidal shape that it resulted in within the field pattern. After exploring the effects of changing the diameter of the center hole, it was decided that the design may benefit from reducing the area of the backplate by as much as possible. The idea behind this being that it would reduce the effect of eddy currents even further. These designs may be seen below in Figure 18. Figure 18 Minimal surface area design This design was pushed as far as possible, resulting in the backplate shown in Figure 19. It was found from these changes that the removal of the wires along the radius of the exciters appeared to decrease the magnitude of the field pattern. Additionally, further changes to this design found that the field pattern may still resemble what it was before these changes, even if there is only a central wire and an outer wire. Overall, this design and its field pattern for this backplate design are shown in Figure 19. Figure 19 Final surface area-minimized design 21

23 Figure 20 Axial magnetic field plot for final surface area-minimized design After the team was confident with this design, referred to as the mesh backplate, it was simulated using ANSYS Maxwell, a software used for modeling low-frequency fields. Although it was found that the eddy currents in this backplate were reduced (by approximated 30-40% according to a surface integral in the ANSYS fields calculator), they were not as low as desired. Images showing a comparison between the eddy currents in the solid backplate and the new backplate are shown below in Figure 21. A 1D plot, showing a comparison between the eddy currents along the radius of the original and mesh backplates, may also be seen below in Figure 21. The red lines show a simulation performed at 5kHz, while the black lines show a simulation for 60Hz, illustrating how the eddy currents increase as frequency increases. The eddy currents, seen especially along the innermost ring, were determined to have been caused by the closed loops within the backplate, so steps were taken to eliminate any closed loops in future designs. Figure 21 1D plot comparing eddy currents along a radial cut of original and mesh backplate The mesh backplate was not able to effectively suppress the eddy currents, so new designs were devised. One such design, shown below in Figure 22 (left), has notches cut along the outermost ring. However, this particular design was found to reduce the field magnitude, so it was not used. Another design, shown in Figure 22 (right), was also tried. This particular design is based off of reflectors for radio towers. Here is can be seen that all of the loops were removed and it was 22

24 found that the eddy currents were reduced dramatically. It also had no negative effects upon the field magnitude. The problem with this design was that it increased the diameter of the backplate, which would have prevented its use in an MRI scanner. This particular backplate was also reduced in size, as shown in Figure 23, but it was found that the azimuthal wires within it, when along the position of the exciters, could not replicate the desired field pattern. Several other variations of this backplate, such as the one seen in Figure 24, were also investigated. Eventually, a final backplate design was found, which will be discussed in the next section. Figure 22 Backplate notches cut along outermost ring (left) and then loops removed (right) Figure 23 Backplate with removal of loops and reduced size Figure 24 Backplate with removal of loops and less connections around the perimeter 23

25 4.2 Final Backplate Design The final backplate, referred to as the sun backplate, due to its sun-like shape, may be seen in Figure 25. As shown, this is similar to the backplates discussed previously, but its diameter does not extend beyond the limit of 36cm and there is a conductive strip along sections of the perimeter of the backplate, which is necessary for its functioning. It was found that the field produced by this backplate was not reduced, despite the lack of wire loops. The 1D field patterns, compared to those produced by the solid backplate, may be seen in Figure 26. This backplate was able to suppress eddy currents as effectively as similar designs discussed in the previous section. When simulated in the ANSYS Maxwell, a surface integral of the current magnitude, along a slice of the z-plane origin, found an approximate 90% reduction in eddy currents in relation to the solid backplate. A volume integral of the current magnitude, over a geometry containing the entire backplate, found that eddy currents were negligible in relation to the solid backplate. Figure 25 "Sun" backplate Figure 26 1D axial magnetic field pattern solid backplate (dashed) and sun backplate (solid) 24

26 Figure 27 2D Field, HRCP plot of the final backplate design There are several important characteristics about the new backplate. First, as shown above, this backplate appears to increase the field magnitude more than the original backplate. Secondly, increasing the inner strip length appears to change the field magnitude on the side of the phantom closest to the backplate, as shown in Figure 28. This also appears to increase the field magnitude and cause a shift in the field, as two of the sinusoidal peaks have been translated to the right, along the z-axis. Reducing the strip length enough will decrease the field magnitude dramatically. This is similar to the phenomenon discovered earlier, where changing the hole size within previous backplate designs would alter the field pattern. Lastly, changing the total number of inner strips appears to affect the field within the phantom. When the number of strips were reduced, such that the total along each section was changed from seven to five, the field magnitude dropped significantly. The angle between each strip was kept equal and would be changed when reducing the total number of strips, such that the angle was dependent upon the number of strips. They were always connected in four sections as well. A more thorough investigation into this backplate is recommended, as it could potentially be altered to increase the field magnitude even further. Figure 28 1D axial magnetic field plot of solid backplate (dashed) vs. sun backplate with thicker strips (solid) 25

27 4.3. Variations in Phantom Distance One consideration in the design is the location of the subject within the coil. Tests were run to determine if the location of the phantom will impact the intensity of the magnetic field. The results are shown below: The trial consistently tested a 60 cm, length and diameter helical coil. The only variable in this testing is the location of the phantom with regards to the z-axis. The phantom stayed on the z- axis and is therefore equal distances between both the x and y-axis. The symmetry is ideal for producing the most uniform magnetic field. The first cylinder location is (0,0,600), in other words, the far edge of the phantom is 600mm from the backplate. This location shows one extreme; the edge of the phantom is right at the end of the coil, furthest from the backplate. Results indicated that the intensity of the magnetic field was averaged about.5 and that from about 125mm to 250mm, the intensity drops to nearly 0. Figure 29 Phantom 815 mm from bakplate (default) Figure 30 Axial magnetic field for phantom 815mm from backplate The next location of the phantom was done at the other extreme, against the back plate. The location of the far end of this phantom is (0, 0, 586). The results below indicate that there is very little change in the intensity as seen in Figure 30 from that of the intensity seen in Figure

28 Figure 31 Phantom 586 mm from backplate Figure 32 Axial magnetic field for phantom 586 mm from backplate The third test was finding the medium between both extremes. This time the phantom is located at (0, 0, 725). These results were the best as far as showing the highest average intensity of the magnetic field. The peak occurred at 55mm and reaches about B1 =.25. The basic shape of this phantom location and the phantom located at (0, 0, 815) are the same but the intensity of the more centrally located phantom has a higher intensity. Figure 33 Phantom 725mm from backplate 27

29 Figure 34 Axial magnetic field for phantom 725mm from backplate 4.4. Variations in Phantom Shape It was believed that the homogeneity and magnitude of the field pattern within the phantom could be manipulated by changing the phantom shape. This would correspond to a making the patient wear some additional accessories during the scan. Although numerous attempts at doing this proved unreliable, a reliable method of manipulating the field pattern within the phantom was eventually discovered. This could be done through the introduction of concave notches or through convex ridges along the side of the phantom. These effects of this technique were replicated in all simulations where it was used. In particular, the introduction of concave notches to the phantom would decrease the field magnitude between the notches, while convex ridges around the side of the phantom would reliably increase the field between the ridges. Using this method, it is possible to correct any inhomogeneity within the field. Shown in Figure 38 it can be seen that the introduction of concave notches makes the field along the z-axis more homogenous. Furthermore, Figure 36 shows that the introduction of convex ridges increases the field between them. This method was simulated for numerous patterns. It was also found that the introduction of convex ridges was less reliable than the method of introducing concave notches. Through further refinement of these notches, it may be possible to make the field pattern along the z-axis more homogenous. This method of introducing notches and ridges around the side of the phantom is flawed, because they are unnatural shapes (i.e. not smooth). For this reason, this method was explored for smooth shapes. By introducing a phantom with a sinusoidal shaped exterior, it was found that this method was still effective, as the field pattern along the z-axis is more homogenous than the field pattern for a phantom without this change. The results of the simulations for this change may be seen in Figure 40. Overall, the reliability of this method makes it worth exploring further. 28

30 Figure 35 Phantom with convex ridges Figure 36 1D axial magnetic field for convex phantom Figure 37 Phantom with concave notches Figure 38 1D axial magnetic field for concave phantom Figure 39 Phantom with sinusoidal shape Figure 40 1D axial magnetic field for sinusoidal phantom 29

31 4.5. Variations in Helix Diameter As previously mentioned, the RF coil can be used for a range of body parts. Because the subject has to fit into the coil, the diameter is something that must be taken into consideration. Simulations were run to determine the effects of a larger coil. The results are shown below: The initial design is a 60cm long, 36 cm radius. The phantom has an 80mm diameter. This small radius showed good results with a high and relatively uniform intensity plot. Figure 41 60cm long, 36cm diameter coil Figure 42 1D axial magnetic field for alternate coil size The first time the diameter increased, it was expanded to a 40cm diameter. The results indicate only a slight decrease in the intensity plot. Both the HRCP and HLCP plots reveal the same pattern. 30

32 Figure cm diamter coil Figure 44 1D axial magnetic field for 40 cm diameter coil 4.6. Designs Involving Straight Wire Antennas The applications of helical antennas with very low pitch were investigated. This investigation eventually found that straight wire antennas could produce field patterns with higher and more even field magnitudes along the center of the phantom. Results of a simulation for a backplate with four monopoles may be seen in Figure 45. This field magnitude could be altered by varying the wire length and also by including multiple backplates. Including another backplate, with a similar configuration, was found to increase the field magnitude even further, which may be seen in Figure 46. Both of these designs use the same power. These suggest that the use of phased arrays and dipoles should be investigated. 31

33 Figure 45 Straight-wire, single backplate model (left) and axial magnetic field (right) Figure 46 Straight-wire, double backplate model (left) and axial magnetic field (right) 32

34 Figure 47 2D field plot of HRCP (left) and HLCP (right) for straight wire design 4.7. Designs Involving Multiple Helices In the attempt to produce a more even field pattern, multiple helices were used. They were then arranged in various ways. One such way may be seen in Figure 48. In this configuration, multiple helical antennas were cascaded, taking advantage of the hole within the backplate. It was found from this arrangement that the field pattern between each backplate hole could be changed. However, this approach was abandoned, as it could not be used to achieve an even field pattern. 33

35 Figure 48 Multiple helix cascade (left) and 2D HRCP plot (right) The approach discussed above was attempted again, except for segments of helices without backplates. With this approach, the feed current for each helix segment could be adjusted, which resulted in field patterns with varied magnitudes. Although the pattern would generally remain the same, the way in which the field was distributed throughout the phantom would change for different combinations of feed currents. It was also found, when adjusting the phase difference, between each helix segment, that this would also change the field distribution. It is possible that changing the phase could potentially change the field pattern, but this requires further investigation. This segmented design was the same one discussed in section 2.10, where data was imported into and manipulated with MATLAB. Other approaches were tried as well, where helices were faced towards one another, which may be seen in Figure 49. The first of these designs involved interweaving two helices, which only worked if one helix was phased in the opposite direction. Another approach did not interweave helices. It was found that the field pattern could be made symmetric if the phantom was placed in the center of these helices, as shown below. Neither of these designs were found to change the field pattern or have any significant advantages. The designs involving monopoles were found to benefit from having multiple antennas. 34

36 Figure 49 Other multi-helix designs 35

37 4.8. Coil Fabrication Upon the completion of optimization, a single design was translated into a physical prototype that will be tested. This required the purchasing of materials and a period of construction. At this point, the design of one coil has been finalized, but there will be additional designs, such as the birdcage coil discussed previously. There were a few obstacle that were found during the design given that the materials allowed in the MRI are limited. The old design was too large and heavy. The reduction of weight came from the replacement of the PVC pipe with styrofoam and vinyl. The process is described below in detail. The construction of the final design took about 2 weeks. The final design consist of the following: 2 circular 12 diameter styrofoam disk 24 long, cylindrical styrofoam pillars 4 rolls of ¼ inch wide copper tape Wooden skewers 5 of vinyl with a thickness of 20mm Cardboard support 4 connectors Acquiring Materials: The styrofoam disks and pillars and wooden skewers were from a local Hobby Lobby. The copper tape was pre-purchased and available in the lab. The cardboard support was found at the local school store. The vinyl was purchased online and required pre planning as it took over two weeks to arrive. Figure 50 Fabricated coil with labelled materials 36

38 The process for construction is as follows: 1. The frame: The styrofoam throughout the project supports the cylindrical form of the coil. The two ends are 12 circular plates held together by 8 evenly spaced, 24 long pillars. The wooden skewers were placed in the ends of each pillar to attach the disks. In addition they were placed between the pillars for additional support. 2. Before wrapping the vinyl, while it was still flat, the lines were drawn to indicate where the copper tape will eventually be placed. Based on calculations, the coil had 5 turns. Marks were placed 12 cm apart on horizontally and 3 cm apart vertically. By connecting the lines diagonally and then wrapping the vinyl, the helical form is marked for the copper taping. 3. The next step was to wrap the marked vinyl around the frame. The vinyl was found to be more flexible than initially anticipated so it was wrapped twice. The vinyl was taped at the end with packaging tape to hold it in place. 4. The copper tape is now laid down over the supportive vinyl. By using one roll of copper tape for each arm of the coil, soldering is not a necessary step. 5. The backplate was constructed last. A cardboard frame was marked similarly to the vinyl to mark where the copper was to be laid down. Because the backplate could not be build with one solid copper strip, once the copper was placed on the cardboard it needed to be soldered. 6. The final step is the connectors. Attaching the connectors was done by drilling a small hole through the cardboard and to the copper tape of the helix. This also involved soldering the ends of the connectors to the copper backplate and helix Hardware Testing After the coil was fabricated, it was ready to be tested. Initially, the coil was planned to be tested at the anechoic chamber on CSU campus using a Precision Network Analyzer (PNA). This lab was only capable of testing the S-parameters. However, with the help of Broadcom and EiR Dr. Osvaldo Buccafusca, we were able to obtain both the S-parameters of each of the helices as well as a relative map of the field inside the coil. 37

39 Figure 51 HFSS simulation (top) and Broadcom experiment (bottom) for 1-channel excitation Figure 52 Smith chart for a single channel The testing results were not all as expected but will improve the project design moving forward. The above Smith Chart shows the impedance of one of the channels of the coil as frequency is swept from 290 to 310 MHz. It was found that the reactance was slightly capacitive when ideally it should be zero (if not slightly inductive). Additionally, the resistance was much less than the desired 50 Ohms required to interface to RF test equipment. With the information given by the Smith chart, the coil can now be matched to 50 Ohms with the addition of small pieces of copper. 38

40 5. Conclusions and Future Work 5.1. Conclusion This project has investigated new RF coil designs for use in next generation MRI scanners. So far, the team has manufactured a prototype of the helical coil design with a modified backplate that should result in the elimination of eddy currents during future tests. Furthermore, this work has also identified reliable methods for improving the performance of RF coils and numerous phenomena which occur by changing current design parameters Further Testing The final stage of testing is at the Center for Magnetic Resonance Research (CMRR) in Minneapolis, MN. It is at this time that the RF coil design will be tested as a part of the whole MRI machine as it acquires an image of a saline phantom. This is the only test capable of observing the effects of eddy currents on the design Recommendation for Continuation The Next-Generation MRI Scanner project is recommended for continuation in the future. Not only does the broader goal of improving MRI technology have massive potential, but also the current project that this team worked on, the 7.0T RF coil, also has some work that needs to be done. The immediate goals of the next team would be to make a few modifications the most recent version of the 7.0T coil. First off, the coil still needs to be matched to the 50 Ohm impedance of the RF sources such that absolute field maps can be obtained (as opposed to what has been obtained already, which are all relative maps). This will also allow the power coupling (an important specification for commercial MRI scanners) to be measured. Also, the RF coil design could potentially be refabricated altogether with better, sturdier (but still non-magnetic and not conductive) materials and more compact connectors. Once these changes have been made, arrangements should be made to test the coil as part of a complete 7.0T MRI system with one of our collaborators, either the CMRR at the University of Minnesota or at the University of Alberta. In addition to work on the current design, there are several other topics that a future team could investigate: Try different RF coil design for the 3.0T, other than a 4-channel helix o Multi-helix, Helmholtz-like configuration o Repeated dipole o 8-channel coil o Coil designs with metamaterials (ambitious) Design coils for 3.0T or 10.5T MRI scanners Design wearable (receiving) coils 39